Fact-checked by Grok 2 weeks ago

Amazon Rekognition

Amazon Rekognition is a cloud-based service offered by (AWS) that employs to enable analysis of images, video streams, and stored videos, facilitating tasks such as , object and scene labeling, text extraction, and . Launched as part of AWS's expansion into tools, Rekognition allows developers to integrate capabilities like identifying facial attributes (e.g., age range, emotions, eyewear), detecting unsafe content, and recognizing celebrities or custom-trained labels without requiring deep expertise in . The service processes vast volumes of media scalably, delivering results with confidence scores in seconds, and supports real-time analysis for applications in , media, and . Its adoption has spanned industries, including for suspect identification and enterprises for automated moderation, underscoring its role in advancing accessible AI-driven . However, Rekognition has drawn significant scrutiny for potential biases in facial recognition accuracy, with empirical tests revealing higher false positive and false negative rates for women and individuals with darker skin tones compared to lighter-skinned males. A notable 2018 test by the (ACLU), using Rekognition to match congressional photos against mugshots, produced 28 false matches—disproportionately affecting people of color—and highlighted error rates exceeding 30% in some demographic categories when thresholds were adjusted below AWS defaults. contested the test's , arguing it employed an overly permissive confidence threshold (below the recommended 90-99%) that inflated false positives, and emphasized ongoing improvements to mitigate disparities observed in benchmarks. These concerns fueled broader debates on , , and algorithmic fairness, culminating in 's 2020 one-year moratorium on sales to U.S. departments amid regulatory pressures and ethical critiques from advocacy groups. Despite such controversies, independent evaluations, including those assessing commercial systems under varied conditions, affirm Rekognition's competitive performance in controlled scenarios while underscoring persistent challenges in equitable accuracy across diverse populations.

Technical Capabilities

Image and Video Analysis Features

Amazon Rekognition employs models to perform automated analysis on and videos, identifying visual elements such as objects, scenes, activities, and text without requiring expertise from users. For , the service detects labels representing thousands of object categories and scene types, returning confidence scores and bounding boxes for precise localization. It also supports (OCR) to extract text from , including printed and handwritten , facilitating applications like . Additionally, Rekognition Image moderates by detecting inappropriate or unsafe elements, such as explicit material, with customizable thresholds for filtering. In video analysis, Rekognition processes both stored videos and streams, delivering frame-accurate results with SMPTE timecodes for temporal precision in workflows. Stored video analysis identifies objects, scenes, and activities across frames, tracking changes over time and generating for segments containing specific elements. The service handles videos up to specified durations, scaling to process large volumes efficiently via AWS infrastructure. Text detection in videos extends OCR to dynamic content, while scans for unsafe visuals throughout the footage. Key analysis operations include:
  • Label detection: Classifies visual content into hierarchical labels (e.g., "person" under "human" or "car" under "vehicle"), applicable to both static images and video segments.
  • Activity recognition: In videos, identifies human activities like "running" or "dancing" with timestamps for event-based querying.
  • Scene understanding: Detects environmental contexts, such as "beach" or "office," aiding in semantic search and categorization.
These features integrate via , returning outputs with metadata for downstream applications, emphasizing scalability for high-throughput analysis.

Facial Recognition and Detection

Amazon Rekognition detects faces in still images and video streams using algorithms, providing bounding boxes around detected faces along with confidence scores indicating the probability of accurate detection. The service identifies up to 100 faces per image, returning positional data such as facial landmarks (e.g., locations of eyes, , and ) to enable precise mapping and orientation assessment. Detection operates in real-time for video analysis, supporting frame-by-frame processing to track faces across sequences. For facial analysis, Rekognition evaluates attributes including estimated age range (e.g., "20-30"), gender classification (male or female), and emotional states such as happiness, sadness, anger, fear, disgust, surprise, or neutral, each with associated confidence levels. Additional attributes cover presence (e.g., , ), detection, and face quality metrics to filter low-confidence results, such as those affected by or poor lighting. These features rely on convolutional neural networks trained on diverse datasets, though performance can vary with image quality, pose, and demographic factors, as evidenced by AWS-reported improvements in detection accuracy over time. Facial recognition extends detection to comparison and identification via APIs like CompareFaces and SearchFaces, which compute similarity scores between face vectors derived from images, enabling 1:1 verification or 1:N searches against collections of up to 100 million faces. Collections store encrypted face metadata for scalable matching, with similarity thresholds adjustable for precision-recall trade-offs; for instance, higher thresholds reduce false positives. In June 2023, AWS introduced user vectors, allowing multiple reference images per identity to boost match accuracy by averaging embeddings, particularly for varied poses or lighting. To counter spoofing, Rekognition Face Liveness analyzes short selfie videos for liveness cues like head movements or blinks, distinguishing real users from photos, masks, or digital replays with reported false acceptance rates under 0.2% in controlled tests. July 2025 updates enhanced liveness accuracy through refined challenge settings and model training, reducing errors in diverse conditions. Overall, while AWS claims high reliability via ongoing model retraining, independent evaluations highlight potential demographic disparities in error rates, underscoring the need for application-specific validation.

Customizable Algorithms

Amazon Rekognition Custom Labels allows users to build tailored models for identifying domain-specific objects, scenes, logos, and concepts in images, extending beyond the service's pre-trained capabilities. Introduced in general availability on December 3, 2019, this feature employs (AutoML) processes to handle algorithm selection, hyperparameter tuning, and model training, enabling deployment without specialized expertise in . To create a custom model, users initiate a project via the AWS Management Console or , import datasets from containing typically a few hundred images, and annotate them with labels or bounding boxes for object localization—either manually in the console or through integration with for scalable labeling. The system automatically partitions the dataset into training (80%) and validation (20%) subsets, trains candidate models in parallel over several hours, and evaluates them using metrics including , , F1 score, and average , selecting the optimal version for production. Deployed models support through the DetectCustomLabels operation, which processes input images and returns detected labels with confidence scores (ranging from 0 to 100) and optional bounding box coordinates for precise localization. incurs costs based on image volume and duration, with no upfront fees, and models can be iterated upon by incorporating additional data or feedback to refine performance. Limitations include restriction to static images (no video or for custom labels), exclusion of , text detection, or tasks (handled by base Rekognition features), and a minimum size to ensure viable .

Development and History

Launch and Initial Release (2017)

Amazon Rekognition, a deep learning-based image and video analysis service, was initially released on February 9, 2017, as part of Amazon Web Services (AWS), enabling developers to integrate computer vision capabilities into applications without managing underlying infrastructure. The service's core image analysis features at launch included object and scene detection, identifying elements such as flowers, coffee tables, and chairs within images; facial detection and analysis, which assessed attributes like emotions, age range, smiling, eyeglasses, and gender; and celebrity recognition, matching faces against a database of public figures. These functionalities were powered by convolutional neural networks trained on vast datasets, allowing for scalable processing via API calls to stored images. Initial availability was limited to the , , and AWS regions, with pricing structured on a pay-per-use basis at $0.001 per image analyzed. The release followed a preview announcement in late 2016, marking the general availability of the image service and accompanying developer guide. Later in 2017, on November 29, AWS expanded Rekognition with video analysis capabilities, supporting object tracking, activity detection, and facial recognition in stored or streaming videos from buckets, further broadening its initial scope for dynamic content processing. These enhancements included real-time face search across large face collections and text detection in images, announced on November 21.

Key Updates and Expansions (2018-2020)

In November 2018, Amazon Rekognition introduced celebrity recognition capabilities, enabling the service to identify hundreds of thousands of celebrities from images and videos by comparing them against a database of public figures. On November 21, 2018, AWS announced enhancements to and analysis, including the ability to detect 40% more faces in challenging images, improved accuracy in face matching, and refined age range estimates with narrower confidence intervals. These updates stemmed from iterative model training on diverse datasets, aiming to reduce false negatives in low-quality or occluded face scenarios. In March 2019, Amazon Rekognition launched its fifth major model update for face analysis, boosting overall accuracy in detection, attribute estimation (such as emotions and landmarks), and recognition tasks across images and videos. This was followed in August 2019 by further improvements to face analysis features, enhancing precision in identifying facial attributes like smiling, eyeglasses, and estimation while maintaining for large-scale applications. By December 2019, AWS added support for text detection in videos, including filters to specify languages and regions, extending the service's utility beyond static images to dynamic content analysis. A significant expansion occurred on November 25, 2019, with the launch of Amazon Rekognition Custom Labels, allowing users to train custom models for detecting specific objects, scenes, or labels without requiring labeled data expertise or deep ML knowledge. This feature automated much of the model training process, enabling domain-specific applications like identifying unique industrial defects or branded items, and was officially released on December 3, 2019. In 2020, enhancements included the addition of an EyeDirection attribute to DetectFaces and IndexFaces operations, providing yaw and predictions for to support advanced behavioral analysis in and contexts. These developments collectively expanded Rekognition's scope from general-purpose recognition to customizable, video-inclusive, and attribute-rich analysis, with ongoing model refinements addressing performance in varied real-world conditions.

Recent Developments and Improvements (2021-2025)

In 2021, Amazon Rekognition introduced enhancements for video analysis, including the ability to detect black frames and primary program content using the StartSegmentDetection and GetSegmentDetection s on June 7. On July 16, customers gained access to a complete list of supported labels and object bounding boxes, enabling better integration and customization of detection workflows. Later that year, on November 11, the DetectLabels was updated to include label aliases, categories, and dominant color detection, improving the granularity of image analysis outputs. By 2022, streaming video capabilities expanded on April 28 with label detection in live streams via ConnectedHome settings, facilitating real-time applications in smart home and scenarios. In November and December, further refinements to label detection APIs added support for aliases, categories, and advanced filtering in both DetectLabels and GetLabelDetection, enhancing precision in scene and object identification. Key advancements in 2023 included the launch of face liveness detection for videos on April 11, which verifies physical user presence to counter spoofing attempts through biometric challenges. On May 9, the model was upgraded for superior detection of explicit and violent material, expanding coverage and reliability. brought bulk image analysis via manifest files in StartMediaAnalysisJob, streamlining processing of large-scale image datasets. Additionally, support for Custom Moderation was introduced around October 12, allowing adapters to refine moderation accuracy for domain-specific content. In 2024, received further enhancements on February 1, incorporating new labels and detection for animated content to broaden applicability in review. A detailed methods to boost Face Search accuracy using user vectors, which embed additional facial embeddings to elevate similarity scores for verified matches. Independent evaluations in a February 2025 arXiv preprint noted improved overall accuracy in Amazon Rekognition compared to prior benchmarks during a 2024 assessment. By mid-2025, Face Liveness detection saw accuracy upgrades and a new FaceMovementChallenge setting on July 3, reducing verification time by approximately 3 seconds while enhancing fraud resistance across challenge modes. These iterative updates have focused on expanding detection scopes, refining outputs, and bolstering anti-spoofing measures, with Custom Moderation extensions announced in June to further tailor moderation precision.

Applications and Adoption

Commercial and Media Sector Uses

In the commercial sector, Amazon Rekognition supports applications such as , identity verification, and across industries including , , and . For instance, company integrates Rekognition to automate returns processing by analyzing product images against purchase records, resulting in 15 times faster handling and higher accuracy compared to manual methods. Similarly, online art marketplace Artfinder employs it for image-based recommendations, enabling of matching algorithms in hours and full production deployment in a week. In , uses the API to scan listing images for compliance, flagging inappropriate content to improve discoverability and . REA Group, another firm, deployed Rekognition's text and label detection in June 2020 to automate compliance checks on property photos, reducing manual review needs and ensuring adherence to platform standards. Financial services leverage Rekognition for secure identity verification. Aella Credit, operating in emerging markets, applies face comparison to loan applications, minimizing verification errors and risks. Q5id achieves a false acceptance rate of 1 in 933 billion through Rekognition-powered biometric checks for high-security . In aviation maintenance, Nordic Aviation Capital implemented custom labels in April 2022 to scan records for defects, automating identification of issues like scratches or and yielding annual savings of up to €200,000. In the media and entertainment sector, Rekognition enables automated video and image analysis for indexing, tagging, and search optimization, reducing reliance on manual labor. utilizes Rekognition Video to detect markers such as shot changes, black frames, and timecodes with frame-accurate SMPTE , saving hundreds of work-hours per year in workflows. Media applies custom labels for tagging of game footage, accelerating content search and retrieval for broadcasters. automates celebrity recognition across digital assets, eliminating repetitive manual tagging for entertainment content libraries. employs facial recognition to index over 7,500 hours of annual video content, more than doubling prior capacity from 3,500 hours. Media technology provider Veritone incorporates Rekognition into its video search pipeline as of May 2024, using shot detection, label generation, text extraction, and celebrity identification to produce granular ; this shot-level indexing boosts search recall by at least 50% over video-level approaches, with further gains of 52% when combined with semantic embeddings. The service's celebrity recognition feature, covering tens of thousands of personalities across categories like and , further streamlines searchable archives by automating identification in images and videos.

Government and Law Enforcement Integration

Amazon Rekognition has been integrated into workflows primarily through its capabilities for analyzing images and videos from cameras, body-worn devices, and databases to detect faces, compare identities, and identify objects or activities relevant to investigations. The service supports scenarios such as matching suspects against mugshot databases, locating missing persons, and aiding in rescues by processing footage to flag potential matches. AWS documentation emphasizes that outputs should serve as investigative leads rather than standalone evidence, requiring human verification to mitigate errors. Early integrations included pilot programs by U.S. departments. In 2019, the Sheriff's Office in deployed Rekognition to scan surveillance videos against booking photos, reportedly identifying suspects in cases like and within hours. The tested real-time facial recognition integration with over 100 officers' body cameras and patrol vehicles in 2018, scanning against a database of 50,000 prior arrests, though the trial concluded without full adoption. These efforts demonstrated potential for rapid suspect identification but raised operational concerns, such as dependency on video quality and database completeness. In June 2020, Amazon imposed a one-year moratorium on sales of Rekognition to U.S. police departments, citing the need for federal regulations on facial recognition amid concerns over misuse and bias. The pause excluded non-law-enforcement entities like nonprofits combating child exploitation, which continued using the tool to assist recoveries in collaboration with authorities. Following the moratorium's end in 2021, Amazon maintained availability for government users with strict guidelines, recommending a minimum 99% confidence threshold for matches in law enforcement contexts and prohibiting sole reliance on the technology for arrests or decisions. A 2024 U.S. Department of Justice disclosure revealed an FBI project in the initiation phase employing Rekognition for image and video analysis, which Amazon stated complied with its policies as it did not constitute direct sales violating the extended commitments against police use. As of 2025, direct adoptions remain limited by policy and scrutiny, with Amazon focusing on indirect integrations via partners for broader enhancements, though critics argue the moratorium's has been inconsistent, allowing persistent through or intermediary channels. Orlando's ongoing initiatives have referenced Amazon partnerships, incorporating high-threshold Rekognition scans into command centers for event monitoring. Overall, integration prioritizes augmenting human-led investigations, with AWS providing tools for custom model training on agency-specific datasets to improve relevance.

Performance and Evaluation

Accuracy Benchmarks and Studies

Amazon has not submitted its Rekognition algorithms for evaluation in the National Institute of Standards and Technology (NIST) Face Recognition Vendor Test (FRVT), the primary independent benchmark for facial recognition accuracy and demographic differentials, as of evaluations through 2020 and no subsequent public participation identified. This absence limits direct comparisons to leading commercial systems, which NIST tests show achieving false non-match rates (FNMR) below 0.1% in controlled 1:1 tasks by 2020, with ongoing improvements to over 99% accuracy in large-scale . AWS maintains that Rekognition delivers "highly accurate" results through continuous model updates incorporating recent and diverse , though specific quantitative benchmarks are not publicly detailed beyond general claims of superior in internal testing. Early independent assessments highlighted variable accuracy. In a 2018 test by the (ACLU), Rekognition compared images of 67 U.S. Congress members against a database of 25,000 mugshots, yielding 28 false matches at default confidence thresholds, with errors disproportionately affecting people of color; countered that proper threshold adjustments (e.g., 99% confidence) reduced false positives to near zero in replications. A 2019 ACLU follow-up using athlete photos reported error rates up to 20.8% for darker-skinned women versus lower for white men, again at lower thresholds. Academic scrutiny in 2019 by researchers found Rekognition exhibited higher accuracy for lighter-skinned faces in gender classification tasks, with error rates increasing for darker skin tones, prompting AWS to dispute the findings as not reflective of real-world configurations emphasizing high-confidence matches. A 2025 re-evaluating commercial APIs, including Rekognition, in a 2024 rerun noted improved accuracy over prior tests (2018-2020), aligning with broader field advancements where systems achieve sub-1% error rates in controlled settings, though specific FNMR figures for Rekognition were not isolated beyond qualitative enhancement. These studies underscore that accuracy depends heavily on operational parameters like thresholds and , with AWS recommending thresholds above 95% to prioritize over in high-stakes applications.

Bias Analysis and Mitigation Efforts

Early evaluations of Amazon Rekognition identified demographic biases in facial analysis tasks, particularly gender classification, with error rates reaching 34.7% for dark-skinned women compared to 0.8% for light-skinned men in a 2018 study using controlled datasets of public figures. A separate 2018 test by the (ACLU) on face matching reported false positive rates up to 5% at default thresholds when comparing photos of U.S. Congress members to mugshots, with nearly 40% of errors involving people of color despite their underrepresentation in the dataset; however, this test used uncalibrated thresholds and uncontrolled image quality, factors Amazon contested as inflating discrepancies. Amazon responded that internal benchmarks on diverse, high-quality datasets showed lower overall errors and smaller differentials, attributing reported biases to study methodologies rather than inherent model flaws, such as inadequate similarity thresholds or non-representative inputs. Independent third-party evaluations, including those aligned with NIST Face Recognition Vendor Test (FRVT) standards, have since demonstrated reduced demographic differentials in leading commercial systems, with U.S. vendors like exhibiting false non-match rates varying by less than 1% across racial groups in controlled 1:1 matching scenarios. Subsequent model updates from 2019 onward incorporated expanded training data to address imbalances, yielding measurable improvements; a 2024 re-audit of tasks reported enhanced accuracy for Amazon Rekognition relative to 2018 baselines, particularly for underrepresented demographics. Recent iBeta certifications for Rekognition Face Matching, conducted under ISO/IEC 19794-5 standards, confirmed true match rates exceeding 99.97% across six demographic categories (age, , skin tone) at a 95% , indicating in performance under operational conditions. Mitigation efforts by Amazon include curating training datasets with deliberate demographic balance to counteract historical skews in public image corpora toward lighter-skinned individuals, alongside configurable similarity thresholds that users can adjust to minimize disparate false positives—e.g., raising thresholds from 80% to 95% reduces errors across groups without sacrificing utility. Complementary tools like Clarify enable pre- and post-deployment detection via metrics such as demographic parity and equalized odds, while research initiatives from Amazon Science propose unlabeled methods to forecast and debias recognition models during development. These approaches reflect causal factors in —primarily mismatches—prioritizing empirical validation over unadjusted outputs, though critics from organizations argue that real-world deployments amplify residual risks due to variable image conditions.

Controversies and Debates

Privacy and Ethical Concerns

Amazon Rekognition's ability to analyze and match facial images against large databases in has raised significant concerns, as it enables the automated tracking of individuals across and spaces without their explicit or awareness. Critics argue that this facilitates , potentially eroding in everyday activities such as walking in or interacting with commercial systems. In a 2018 demonstration by the (ACLU), Rekognition incorrectly matched 28 members of the U.S. Congress to publicly available mugshots, with disproportionate errors affecting people of color, highlighting the risks of deploying such tools in investigative contexts where expectations are high. Ethically, the technology's integration into law enforcement workflows prompts debates over the delegation of human judgment to algorithms, which may prioritize efficiency over and amplify power imbalances between state actors and citizens. Over 100 academics and researchers signed an in 2018 urging to halt sales of Rekognition to and government entities, citing the potential for authoritarian misuse in suppressing dissent or enabling unchecked . has countered that its service includes safeguards like customer-managed and that, as of early 2019, no misuse reports had been received from users, emphasizing responsible deployment through that prohibit harmful applications. Further ethical scrutiny arises from Rekognition's compatibility with consumer devices, such as doorbells, which could extend facial analysis into residential areas and blur lines between personal security and pervasive monitoring of neighbors or visitors. AWS documentation advises against inputting sensitive into the system to mitigate risks, but concerns persist over how processed biometric information is stored, shared, or protected against unauthorized access by customers or third parties. In response to mounting pressure, Amazon implemented a one-year moratorium on Rekognition sales to U.S. departments in June 2020, though federal agencies like the FBI initiated evaluation projects with the tool by 2024, reigniting discussions on the adequacy of self-regulation versus statutory oversight.

Claims of Demographic Bias and Error Rates

In 2018, researchers from and the published the Gender Shades study, which assessed gender classification accuracy in three commercial facial analysis systems, including Amazon Rekognition, using datasets balanced for skin type and gender. The study reported error rates for Rekognition reaching 34.7% for darker-skinned women ( IV-VI), compared to 0.8% for lighter-skinned men, attributing disparities to underrepresentation of darker-skinned females in training data. The tested Rekognition in July 2018 by comparing images of 67 U.S. Congress members to a database of 25,000 mugshots, yielding 28 false positive matches under default settings. Of these, nearly 40% involved people of color, who represented only 20% of the congressional images, a result the ACLU cited as of racial aligning with prior findings on higher inaccuracies for darker-skinned faces and women. A December 2019 NIST report on 189 algorithms from 99 developers, drawn from 18.27 million images, documented demographic differentials in face recognition performance, with false positive rates in matching 10 to 100 times higher for Asian and African American faces than for faces in many systems, and elevated rates for African American females in one-to-many searches. While did not submit Rekognition to early FRVT phases evaluated in the report, later submissions showed Rekognition exhibiting false positive differentials by race and sex, though less severe than some competitors.
Demographic GroupReported Error Rate (Gender Shades, 2018)
Lighter-skinned males0.8%
Darker-skinned femalesUp to 34.7%
A 2020 re-audit of Gender Shades benchmarks noted reduced error rates for Black females in Rekognition compared to 2018 levels, yet advocates maintained that residual disparities stemmed from biased training datasets lacking sufficient diversity. These claims, often amplified by groups, emphasized risks of misidentification in applications, though NIST analyses indicated that equitable performance across demographics was achievable in some algorithms.

Responses from Amazon and Independent Rebuttals

Amazon has consistently defended the accuracy of Rekognition by emphasizing the importance of adjustable confidence thresholds, arguing that critics often test at inappropriately low similarity scores, leading to inflated false positive rates. In response to a 2018 (ACLU) experiment that reported 28 false matches of U.S. congressional members to mugshots using Rekognition, stated that the ACLU applied a default threshold of 0% similarity rather than the recommended 99% or higher, which would drastically reduce errors in operational use. The company highlighted that no facial recognition system achieves perfect accuracy, but proper threshold calibration aligns Rekognition's performance with industry benchmarks, citing internal evaluations showing over 99% accuracy on large datasets when thresholds are set correctly. Addressing a January 2019 study by researchers and , which claimed higher error rates for darker-skinned females in gender classification, Amazon contended that the paper conflated facial analysis (attribute detection) with facial recognition (identity matching) and used non-standard metrics that did not reflect real-world deployment. Amazon asserted that its internal testing, including on diverse datasets, demonstrated high-quality results contradicting the study's portrayal of "low quality" outputs, and noted ongoing improvements to mitigate any identified disparities. In February 2019, Amazon described broader bias allegations as "misleading," pointing to enhancements like non-binary gender classification options introduced post-study to better handle demographic variations. Independent analyses have partially rebutted Amazon's defenses by underscoring methodological flaws in critics' tests while acknowledging threshold sensitivity, but also confirming persistent demographic differentials. A December 2019 National Institute of Standards and Technology (NIST) evaluation of 189 facial recognition algorithms, including Rekognition, found that while U.S.-developed systems like Amazon's generally outperformed international ones, false positive rates for Asian and African American faces were 10 to 100 times higher than for faces across vendors, even at calibrated thresholds, attributing this to training data imbalances rather than misuse alone. Buolamwini rebutted Amazon's internal-testing claims in January 2019, arguing that self-evaluations lack and external audits are essential to verify -free assertions, as corporate incentives may prioritize over comprehensive demographic . Subsequent tests in NIST's Face Recognition Vendor Test (FRVT) showed Amazon's submissions improving over time, with reduced but non-zero in later iterations, supporting Amazon's improvement narrative while validating critics' calls for standardized, benchmarking.

Policy Restrictions and Moratoriums

In June 2020, Amazon announced a one-year moratorium on sales of Rekognition's facial recognition capabilities to U.S. police departments, citing the need for federal legislation to regulate the technology's use amid concerns over potential misuse. This followed pressure from civil rights groups and researchers highlighting accuracy issues and risks of biased outcomes in law enforcement applications. The moratorium was extended indefinitely in May 2021, with stating it would pause access until governments established appropriate regulatory frameworks. However, has maintained that the policy does not apply to agencies like the FBI, which has continued purchasing Rekognition for image and video analysis, as the company interprets "" to exclude non-local entities. Critics, including the Surveillance Technology Oversight Project, argue this distinction undermines the moratorium's intent and enables unchecked deployment. Several U.S. municipalities have enacted outright bans on government use of facial recognition technology, effectively restricting Rekognition in public sector applications. San Francisco prohibited city agencies from acquiring or using such systems in May 2019, becoming the first major U.S. city to do so. Similar ordinances followed in Oakland, Boston, and Portland, Oregon, by mid-2020, barring police and other agencies from deploying the technology due to privacy risks and error rates. In June 2021, King County, Washington—encompassing Seattle—became the first U.S. county to ban government-wide use of facial recognition, extending prohibitions to non-law enforcement entities. At the state level, restrictions have proliferated, with multiple jurisdictions enacting bans or moratoriums on applications by 2022, often encompassing commercial systems like Rekognition. Federally, proposed legislation such as the Facial Recognition Ban on Body Cameras Act has sought broader prohibitions but has not passed as of 2025. In the , the AI Act—effective from August 2024 with prohibitions phased in from February 2025—classifies real-time biometric identification in public spaces as prohibited for most uses, imposing strict limits on systems like Rekognition unless justified for specific exemptions under narrow conditions. These regulations prioritize data protection under GDPR, requiring high-risk AI providers to demonstrate compliance, though enforcement specifics for cloud-based services remain under development. Amazon Rekognition undergoes third-party audits for compliance with frameworks including reports, PCI DSS, and , enabling its use in regulated environments such as government applications. For federal compliance, it supports validated cryptographic modules via dedicated endpoints, facilitating secure processing in high-security contexts like AWS GovCloud (). AWS documentation emphasizes customer responsibility for data sensitivity and applicable laws, recommending against storing sensitive information in metadata fields, though it does not specify unique Rekognition protocols beyond general AWS measures. Legal scrutiny has centered on allegations of unauthorized biometric data collection under Illinois' Biometric Information Privacy Act (BIPA). In 2022, a federal claiming Amazon violated BIPA by collecting and analyzing facial geometry from images was dismissed, with the court ruling that the plaintiffs failed to establish under the statute. Separate litigation persists regarding Amazon Photos, where plaintiffs assert that Rekognition scans uploaded personal photos to extract and store biometric identifiers like facial embeddings, refining algorithms without user notification or , potentially breaching BIPA's and retention requirements. In March 2023, a lawsuit was filed against for deploying facial recognition in its retail stores without informing customers, alleging violations of local privacy expectations and broader biometric laws, though specific outcomes remain pending. These cases highlight tensions between technological deployment and state-level biometric regulations, with critics arguing that self-reported compliance lacks enforceable oversight, while maintains adherence to legal standards through contractual safeguards and usage guidelines. No federal-level enforcement actions or fines directly targeting Rekognition's core functionality have been reported as of 2025, though disclosures of its integration in FBI projects have prompted calls for greater transparency.

References

  1. [1]
  2. [2]
    What is Amazon Rekognition? - Amazon Rekognition
    Amazon Rekognition is a cloud-based image and video analysis service that makes it easy to add advanced computer vision capabilities to your applications.
  3. [3]
    Amazon Rekognition Customers - Amazon Web Services (AWS)
    Using an already-built solution saved a great deal of time and money in order to make our spring 2018 launch date, and the scalability of cloud-based tools ...
  4. [4]
    Accuracy Comparison Across Face Recognition Algorithms
    ... In another report, Amazon Rekognition had a very low error (< 0.1%) for lighter skinned male populations, with a contrasting 31.4% error rate for darker ...
  5. [5]
    A phenomenological perspective on AI ethical failures: The case of ...
    Apr 1, 2023 · In January 2019, the MIT study extended its investigation on commercial facial recognition programs to Amazon Rekognition. They found that ...
  6. [6]
    Amazon's Face Recognition Falsely Matched 28 Members of ... - ACLU
    Jul 26, 2018 · Our test used Amazon Rekognition to compare images of members of Congress with a database of mugshots. The results included 28 incorrect matches.
  7. [7]
    ACLU Comment on New Amazon Statement Responding to Face ...
    Jul 27, 2018 · The test revealed that Rekognition falsely matched 28 current members of Congress with images in an arrest photo database.
  8. [8]
    ACLU Statement on Amazon Face Recognition Moratorium
    Jun 10, 2020 · ... Amazon's Rekognition software incorrectly matched 28 members of Congress, identifying them as other people who have been arrested for a crime.
  9. [9]
  10. [10]
    Amazon Rekognition Image - AWS
    Rekognition Image is a deep learning powered image recognition service that detects objects, scenes, and faces; extracts text; recognizes celebrities; ...Missing: video | Show results with:video
  11. [11]
    Detecting objects and concepts - Amazon Rekognition
    Amazon Rekognition provides deep learning-based image and video analysis capabilities, including object detection, facial analysis, and unsafe content detection ...Detecting labels in an image · Detecting labels in a video · Detecting custom labels
  12. [12]
    Amazon Rekognition FAQs - AWS
    Rekognition Image also extracts text, recognizes celebrities, and identifies inappropriate content in images. It also allows you to search and compare faces.
  13. [13]
    Amazon Rekognition Video for media analysis - AWS
    Amazon Rekognition Video media analysis features provide frame-accurate detection results along with SMPTE timecodes. This means that you get the exact frame ...
  14. [14]
    Amazon Rekognition Video - AWS
    Amazon Rekognition Video is a fully managed machine learning (ML) service that supports both real time streaming video events and stored video analysis.
  15. [15]
    Understanding Rekognition's image and video operations
    Amazon Rekognition provides deep learning-based image and video analysis capabilities, including object detection, facial analysis, and unsafe content detection ...Amazon Rekognition Image... · Amazon Rekognition Video...
  16. [16]
    Amazon Rekognition Video Documentation - AWS
    Key features for stored video analysis · Object, scene, and activity detection · Content moderation · Text detection · Celebrity recognition · Face detection and ...<|separator|>
  17. [17]
    Working with images and videos - Amazon Rekognition
    Amazon Rekognition provides deep learning-based image and video analysis capabilities, including object detection, facial analysis, and unsafe content detection ...
  18. [18]
    Detecting and analyzing faces - Amazon Rekognition
    Amazon Rekognition can detect faces, identify landmarks, analyze emotions, and other attributes like glasses presence and face occlusion.Comparing faces in images · Detecting faces in an image
  19. [19]
    Amazon Rekognition Announces Real-Time Face Recognition ...
    Nov 21, 2017 · Customers who are already using Amazon Rekognition for face verification and identification will experience up to a 10% accuracy improvement in ...Amazon Rekognition Announces... · Text In Image · Crowd-Mode Face Detection<|separator|>
  20. [20]
    Improve accuracy of Amazon Rekognition Face Search with user ...
    Apr 24, 2024 · In this post, we demonstrate how to use the Amazon Rekognition Face Search APIs with user vectors to increase the similarity score for true matches.
  21. [21]
    Amazon Rekognition Face Liveness - AWS
    Amazon Rekognition Face Liveness analyzes a short selfie video to detect spoofs presented to the camera, such as printed photos, digital photos, digital videos ...Use Cases · Capabilities · General
  22. [22]
    Amazon Rekognition Face Liveness launches accuracy ...
    Jul 3, 2025 · Today, AWS announces accuracy improvements and new settings for Amazon Rekognition Face Liveness. Amazon Rekognition Face Liveness is a ...
  23. [23]
    AWS launches Amazon Rekognition Custom Labels to enable ...
    Dec 3, 2019 · Amazon Rekognition Custom Labels is now generally available in US East (N.Virginia), US East (Ohio), US West (Oregon), and EU (Ireland) regions.
  24. [24]
    What is Amazon Rekognition Custom Labels?
    Amazon Rekognition Custom Labels identifies objects, logos, and scenes specific to your business needs, training a model with a small set of images.Key benefits · Choosing to use Amazon...
  25. [25]
    Getting started with Amazon Rekognition Custom Labels
    Amazon Rekognition Custom Labels trains models to analyze images. Start by choosing an example project, then train, start, and use the model.
  26. [26]
    Training an Amazon Rekognition Custom Labels model
    You can train a model by using the Amazon Rekognition Custom Labels console, or by the Amazon Rekognition Custom Labels API. If model training fails, ...
  27. [27]
    Understanding Amazon Rekognition Custom Labels
    Amazon Rekognition Custom Labels trains models to find objects, scenes, concepts, object locations, or brand locations within image datasets.
  28. [28]
    Detecting custom labels - Amazon Rekognition - AWS Documentation
    Amazon Rekognition Custom Labels can identify the objects and scenes in images that are specific to your business needs, such as logos or engineering ...
  29. [29]
    Amazon Rekognition Custom Labels - AWS
    Amazon Rekognition Custom Labels identifies objects and scenes specific to your business needs, using a small set of training images, and builds off existing  ...
  30. [30]
    Custom train videos in Rekognition | AWS re:Post
    Nov 2, 2021 · Rekognition Custom Labels currently does not support training or inference on videos. Currently the only way to use Rekognition Custom Labels to ...
  31. [31]
    Document history for Amazon Rekognition
    The following table describes important changes in each release of the Amazon Rekognition Developer Guide.
  32. [32]
    Introducing Amazon Rekognition - AWS
    Nov 30, 2016 · Amazon Rekognition is a service that makes it easy to add image analysis to your applications. With Rekognition, you can detect objects, ...Missing: initial | Show results with:initial
  33. [33]
    Image Detection and Recognition Powered by Deep Learning - AWS
    Nov 30, 2016 · Rekognition is now available in the US East (N. Virginia), US West (Oregon), and Europe (Ireland) Regions and you can start using it today. As ...
  34. [34]
    Welcoming Amazon Rekognition Video: Deep-Learning Based ...
    Nov 29, 2017 · It was this time last year during re:Invent 2016 that Jeff announced the Amazon Rekognition service launch. I was so excited about getting ...
  35. [35]
    Amazon Rekognition announces real-time face recognition, Text in ...
    Amazon Rekognition announces real-time face recognition, Text in Image recognition, and improved face detection. Posted on: Nov 21, 2017.
  36. [36]
    Amazon Rekognition announces updates to its face detection ...
    Nov 21, 2018 · With the latest updates, Amazon Rekognition can now detect 40% more faces - that were previously missed - in images that have some of the most ...
  37. [37]
    Amazon Rekognition announces updates to its face detection ...
    Nov 21, 2018 · These updates provide customers with improvements in the ability to detect more faces from images, perform higher accuracy face matches, and ...
  38. [38]
    Amazon Rekognition Launches Enhanced Face Analysis - AWS
    Mar 18, 2019 · Today, we are launching accuracy enhancements to the face analysis features. This is the fifth model update overall since the service launched.
  39. [39]
    Amazon Rekognition improves Face Analysis - AWS
    Aug 12, 2019 · Amazon Rekognition provides a comprehensive set of face detection, analysis, and recognition features for image and video analysis.
  40. [40]
    Amazon Rekognition - AWS API Changes
    Jun 6, 2025 · 2018/08/23 - 1 new 12 updated api methods. Changes This release introduces a new API called DescribeCollection to Amazon Rekognition. You can ...
  41. [41]
    Announcing Amazon Rekognition Custom Labels
    Nov 25, 2019 · Amazon Rekognition Custom Labels will be generally available on December 3, 2019. Click here to be notified when the service becomes available.
  42. [42]
    AWS customers can now view all the labels supported by Amazon ...
    Jul 16, 2021 · Starting today, customers can view the complete list of labels and object bounding boxes supported by Amazon Rekognition, to quickly identify ...
  43. [43]
    [PDF] arXiv:2502.14996v1 [cs.CV] 20 Feb 2025
    Feb 20, 2025 · The. 2024 rerun shows similar results for three out of four services (Face++, Luxand, Tencent) and improved accuracy for Amazon Rekognition. To ...
  44. [44]
    How REA Group implemented automated image compliance ... - AWS
    Jun 9, 2020 · In media and entertainment applications, you can catalog videos based on the text on the screen, such as ads, news, sport scores, and captions.About Rea Group · Image Compliance Solution · How It Works<|separator|>
  45. [45]
    How Nordic Aviation Capital uses Amazon Rekognition to ... - AWS
    Apr 25, 2022 · With Amazon Rekognition Custom Labels, NAC built an AI solution that could automatically scan aircraft maintenance records and identify, based ...Missing: sector | Show results with:sector
  46. [46]
    How Veritone uses Amazon Bedrock, Amazon Rekognition ... - AWS
    May 7, 2024 · In this post, we demonstrate how to use enhanced video search capabilities by enabling semantic retrieval of videos based on text queries.
  47. [47]
    Recognize celebrities in images and videos using Amazon ... - AWS
    Aug 26, 2021 · The celebrity recognition feature in Amazon Rekognition automatically recognizes tens of thousands of well-known personalities in images and videos using ...
  48. [48]
    Use cases that involve public safety - Amazon Rekognition
    Amazon Rekognition can help in public safety and law enforcement scenarios—such as finding lost children, combating human trafficking, or preventing crimes. In ...
  49. [49]
    How Amazon Rekognition helps in the fight against some of the ...
    Jul 9, 2019 · Lebel now uses Amazon Rekognition in his work to identify and locate victims, then collaborate with law enforcement to conduct recoveries.
  50. [50]
    The Facts on Facial Recognition with Artificial Intelligence - AWS
    In all public safety and law enforcement scenarios, technology like Amazon Rekognition should only be used to narrow the field of potential matches. The ...
  51. [51]
    Amazon's facial-recognition AI is supercharging police in Oregon ...
    Apr 30, 2019 · A sheriff's office in Oregon is using Amazon Rekognition to help track down criminal suspects. Is this a dangerous precedent against privacy?
  52. [52]
  53. [53]
    We are implementing a one-year moratorium on police use of ...
    Jun 10, 2020 · We're implementing a one-year moratorium on police use of Amazon's facial recognition technology. We will continue to allow organizations like Thorn, the ...
  54. [54]
    Amazon recommends 99% or higher confidence match when using ...
    We continue to recommend that customers do not use less than 99% confidence levels for law enforcement matches, and then to only use the matches as one input ...
  55. [55]
    Justice Department discloses FBI project with Amazon Rekognition ...
    Jan 25, 2024 · “Amazon Rekognition offers pre-trained and customizable computer vision (CV) capabilities to extract information and insights from lawfully ...
  56. [56]
    Amazon defends facial-recognition tech sale to FBI despite moratorium
    Feb 9, 2024 · The FBI has confirmed it is using Amazon's Rekognition image and video analysis software, but Amazon says it is 'false' to suggest this violates ...
  57. [57]
    Amazon's quiet rise as a power broker in police surveillance
    Oct 6, 2025 · Amazon paused direct sales of Rekognition to police in 2020 after studies showed racial bias, but critics point out that Amazon never fully ...
  58. [58]
    Under Orlando's real-time surveillance partnership with Amazon ...
    Jul 19, 2025 · Amazon now said it recommended law enforcement agencies using Rekognition set the confidence threshold at 99 percent. Wood, though ...
  59. [59]
    Five things you need to know about a gov't study on face surveillance
    Feb 3, 2020 · Amazon's flagship facial recognition software, Rekognition, was not submitted for testing by NIST in parts I, II, or III of the FRVT. Amazon ...
  60. [60]
    [PDF] Questions for Charles H. Romine Director, Information Technology ...
    Jun 4, 2019 · Amazon has not submitted algorithms to NIST FRVT. d. Are companies notified of the results of NIST's findings? NIST Response: Yes. We email ...
  61. [61]
    [PDF] Face Recognition Vendor Test (FRVT) Part 8 - NIST Pages
    That proportion - the failure to acquire rate - can exceed the core recognition error rates2. This report deals with summarizing performance differences of face ...<|separator|>
  62. [62]
    Thoughts on Recent Research Paper and Associated Article ... - AWS
    Jan 26, 2019 · The research paper states that Amazon Rekognition provides low quality facial analysis results. This does not reflect our own extensive testing.
  63. [63]
    Amazon Facial Recognition Falsely Links 27 Athletes to Mugshots in ...
    Oct 28, 2019 · The software is most accurate, the study found, when parsing images of white men. “For darker-skinned women … the error rates were 20.8 ...<|separator|>
  64. [64]
    Amazon facial-identification software used by police falls short on ...
    Jan 25, 2019 · Researchers with M.I.T. Media Lab also said Amazon's Rekognition system performed more accurately when assessing lighter-skinned faces, raising ...
  65. [65]
    About-Face: Examining Amazon's Shifting Story on Facial ...
    Apr 10, 2019 · Explaining how police setting “confidence thresholds” impacts the accuracy of facial recognition systems, and tracking how Amazon has ...
  66. [66]
    Study finds gender and skin-type bias in commercial artificial ...
    Feb 11, 2018 · Three commercially released facial-analysis programs from major technology companies demonstrate both skin-type and gender biases, according to a new paper.
  67. [67]
    The Myth of Facial Recognition Bias - Clearview AI
    Nov 28, 2022 · In the NIST 1:1 FRVT that evaluates demographic accuracy, Clearview AI's algorithm consistently achieved greater than 99% accuracy across all ...Missing: benchmarks | Show results with:benchmarks
  68. [68]
    Amazon Rekognition Face Matching - AWS AI Service Cards
    Amazon Rekognition face matching measures similarity between two face images, using CompareFaces and SearchFaces APIs, and returns a similarity score.
  69. [69]
    A framework to mitigate bias and improve outcomes in the new age ...
    Jun 26, 2023 · AWS offers several capabilities for mitigating bias on your training data. You can use Amazon SageMaker Data Wrangler for feature engineering ...
  70. [70]
    Method predicts bias in face recognition models using unlabeled data
    We hope that this methodology will help AI practitioners working on face recognition or similar biometric tasks to ensure the fairness of their models. Research ...Missing: Rekognition studies
  71. [71]
    Amazon Should Stop Selling a Technology That Powers ...
    ... Amazon to not make Rekognition available to government agencies. “People should be free to walk down the street without being watched by the government.Missing: 2020-2025 | Show results with:2020-2025
  72. [72]
    Facial Recognition: Data Privacy Concerns on Your Own Block
    Amazon provides for a useful case study in examining the potential privacy and civil liberties problems unregulated facial recognition presents. Studies have ...Missing: "peer | Show results with:"peer
  73. [73]
    The two-year fight to stop Amazon from selling face recognition to ...
    Jun 12, 2020 · The results, published half a year later in a peer-reviewed paper, once again found huge technical inaccuracies. Rekognition was classifying ...
  74. [74]
    Open Letter to Amazon against Police and Government use ... - ICRAC
    Researchers in Support of Amazon Employees: Amazon should not sell Face Rekognition Technologies to Police, Government or Intermediaries.
  75. [75]
    Some Thoughts on Facial Recognition Legislation - AWS
    Feb 7, 2019 · Facial recognition technology significantly reduces the amount of time it takes to identify people or objects in photos and video.Missing: studies | Show results with:studies
  76. [76]
    Amazon's Disturbing Plan to Add Face Surveillance to Your Front Door
    like Rekognition, the product that the company ...
  77. [77]
    Data protection in Amazon Rekognition - AWS Documentation
    We strongly recommend that you never put confidential or sensitive information, such as your customers' email addresses, into tags or free-form text fields.Missing: violations | Show results with:violations
  78. [78]
    NIST Study Evaluates Effects of Race, Age, Sex on Face ...
    Dec 19, 2019 · A new NIST study examines how accurately face recognition software tools identify people of varied sex, age and racial background.Missing: Amazon | Show results with:Amazon
  79. [79]
    [PDF] Face Recognition Vendor Test (FRVT) Part 8
    This report summarizes demographic differences in face recognition, analyzing false positive and negative error rates across age, sex, and race.Missing: Rekognition | Show results with:Rekognition
  80. [80]
    Racial Discrimination in Face Recognition Technology
    Oct 24, 2020 · Face recognition algorithms boast high classification accuracy (over 90%), but these outcomes are not universal. A growing body of research ...
  81. [81]
    Thoughts On Machine Learning Accuracy | AWS News Blog
    Jul 27, 2018 · This blog shares some brief thoughts on machine learning accuracy and bias. Let's start with some comments about a recent ACLU blog in which ...
  82. [82]
    Amazon: Facial recognition bias claims are 'misleading' - BBC
    Feb 4, 2019 · While none was 100% accurate, it found that Amazon's Rekognition tool performed the worst when it came to recognising women with darker skin.
  83. [83]
    Many Facial-Recognition Systems Are Biased, Says U.S. Study
    Dec 19, 2019 · Algorithms falsely identified African-American and Asian faces 10 to 100 times more than Caucasian faces, researchers for the National Institute ...
  84. [84]
  85. [85]
    Amazon extends moratorium on police use of facial recognition ...
    May 18, 2021 · Amazon.com Inc (AMZN.O)said on Tuesday it is extending until further notice a moratorium it imposed last year on police use of its facial ...Missing: sales | Show results with:sales
  86. [86]
    Amazon will block police indefinitely from using its facial-recognition ...
    May 18, 2021 · Amazon said it would stop providing its facial-recognition software, Rekognition, to police for a year. Now it appears that ban won't be ending any time soon.
  87. [87]
    S.T.O.P. Condemns Amazon For Breaking Police AI Moratorium ...
    Jan 26, 2024 · The civil rights group renewed its call on Amazon to permanently stop law enforcement use of facial recognition and for legislation outlawing ...
  88. [88]
    San Francisco Bans Facial Recognition Technology
    May 14, 2019 · The ban prohibits city agencies from using facial recognition technology, or information gleaned from external systems that use the technology.
  89. [89]
    13 Cities Where Police Are Banned From Using Facial Recognition ...
    13 Cities Where Police Are Banned From Using Facial Recognition Tech · 1. San Francisco, Calif. · 2. Boston, Mass. · 3. Portland, Ore. · 4. Portland, Maine · 5.Missing: Rekognition | Show results with:Rekognition
  90. [90]
    Amazon bans police use of face recognition technology for a year
    Jun 11, 2020 · A number of U.S. cities have banned its use by police and other government agencies, led by San Francisco last year. On Tuesday, IBM said it ...
  91. [91]
    Seattle's King County becomes first U.S. county to ban government ...
    Jun 1, 2021 · In a unanimous vote, the King County Council approved a ban on law enforcement use of facial recognition technology, making it the first ...<|separator|>
  92. [92]
    [PDF] Amazon markets and sells facial recognition ("Rekognition”) to ...
    Amazon issued an indefinite moratorium on Rekognition used by police departments. ... Microsoft banned face recognition sales to police awaiting federal ...
  93. [93]
    [PDF] Regulating facial recognition in the EU - European Parliament
    In addition to the existing applicable legislation (e.g. data protection and non- discrimination), the draft AI act proposes to introduce new rules governing ...Missing: Rekognition | Show results with:Rekognition
  94. [94]
    Building trust in AI: The AWS approach to the EU AI Act
    Jun 19, 2025 · As the European Commission continues to publish compliance guidance, such as the Guidelines of Prohibited AI Practices and the Guidelines on AI ...
  95. [95]
    Compliance validation for Amazon Rekognition
    Third-party auditors assess the security and compliance of Amazon Rekognition as part of multiple AWS compliance programs. These include SOC, PCI, FedRAMP, ...
  96. [96]
    Using Amazon Rekognition as a FedRAMP authorized service
    To be FedRAMP compliant, you can use a Federal Information Processing Standard (FIPS) endpoint. This gives you access to FIPS 140-2 validated cryptographic ...
  97. [97]
    Amazon Rekognition in AWS GovCloud (US)
    Amazon Rekognition makes it easy to add image and video analysis to your applications. You just provide an image or video to the Rekognition API.
  98. [98]
    Amazon class action over biometric data collection dismissed
    Oct 19, 2022 · Amazon faces a class action lawsuit from consumers who claim the company illegally obtained biometric identifiers for facial recognition.
  99. [99]
    Amazon Photos Biometric Privacy Lawsuit
    The lawsuit alleges that, through its Amazon Photos service, Amazon has for years violated the Biometric Information Privacy Act.Missing: Rekognition | Show results with:Rekognition<|control11|><|separator|>
  100. [100]
    Recent Lawsuits Against AI Companies: Beyond Copyright ...
    Rating 5.0 (16) May 19, 2025 · Plaintiffs allege that Rekognition scans faces from personal photos uploaded to Amazon Photos, uses that data to refine its algorithm, and then ...<|separator|>
  101. [101]
    Amazon sued for not telling New York store customers about facial ...
    Mar 16, 2023 · Amazon did not alert its New York City customers that they were being monitored by facial recognition technology, a lawsuit filed Thursday ...