Sama
Sama (formerly Samasource) provides data annotation and validation services for AI model training, specializing in image, video, 3D point cloud, and text labeling across industries including autonomous vehicles, retail, agriculture, robotics, and financial services.
Profile
Provides human-verified data annotation, validation, and AI model evaluation services for machine learning algorithms.
Sama (formerly Samasource) provides data annotation and validation services for AI model training, specializing in image, video, 3D point cloud, and text labeling across industries including autonomous vehicles, retail, agriculture, robotics, and financial services. Founded in 2008 by Leila Janah as a nonprofit with a social mission—connecting workers in developing regions to fair-wage employment—the company restructured as a for-profit B Corporation in 2019 while maintaining its impact focus. Janah passed away in January 2020, and COO Wendy Gonzalez became CEO.
Sama raised $14.8M in Series A (November 2019, led by Ridge Ventures) and $70M in Series B (November 2021, led by Caisse de dépôt et placement du Québec). The company operates with approximately 4,000+ full-time data experts, holds ISO 9001, 27001, and 42001 certifications, and claims a 99% first-batch acceptance rate. Notable clients include Microsoft, Walmart, eBay, NASA, Sony, Qualcomm, Verizon, Siemens, and Getty Images.
In 2025, Sama launched Bulk Annotation, reducing labeling time by up to 80% and improving consistency by 25%. However, the company faced significant reputational damage: in 2022–2023, reporting revealed that Kenyan workers earning $1.32–$2/hour had labeled toxic content for OpenAI's ChatGPT safety system, experiencing severe trauma; Sama terminated the OpenAI contract early in February 2022. More critically, in early 2026, Meta discovered Sama employees in Kenya had accessed private footage from Ray-Ban Meta AI glasses, triggering contract termination and the layoff of 1,108 workers in Nairobi in April 2026—representing one of Kenya's largest single workforce reductions and underscoring the company's client concentration risk.
Who buys this
- Autonomous vehicle and ADAS companies training perception systems on image and video data
- Retail and e-commerce firms managing catalog annotation and product variation classification
- Robotics and manufacturing companies developing computer vision for industrial automation
- Generative AI developers requiring foundation model evaluation and safety testing
- Tech and media companies building content moderation and classification systems
Publicly disclosed clients
- Microsoft
- Walmart
- eBay
- NASA
- Sony
- Qualcomm
- Verizon
- Siemens
Strengths and what to watch
Strengths
- Achieved 99% first-batch acceptance rate and holds ISO 9001, 27001, and 42001 certifications plus GDPR compliance
- Scaled to 4,000+ full-time data experts (not crowdsourced) with 8-year average customer retention
- Launched Bulk Annotation in 2025, increasing throughput by up to 80% and reducing inconsistencies by 25%
Watch for
- Severe client concentration risk: Meta contract termination resulted in 1,108 layoffs in Kenya (April 2026), wiping out ~25% of reported workforce
- Reputational damage from OpenAI ChatGPT safety work exposure (2022–2023): Kenyan workers earning $1.32–$2/hour reported psychological trauma from toxic content; company terminated contract early under media pressure
- Ray-Ban Meta AI glasses privacy scandal (February 2026): Sama employees accessed intimate footage recorded by users' devices; ongoing UK, Kenya, and California regulatory investigations and class-action lawsuits
Recent moves
Key Information
- Industry
- Data Generation & Labelling
- Founded
- 2008
Frequently Asked Questions
What does Sama do?
Sama provides human-verified data annotation, validation, and AI model evaluation services specifically for machine learning. The company labels images, video, 3D point clouds, and text across diverse industries including autonomous vehicles, retail, robotics, and financial services, serving enterprise clients like Microsoft and Walmart.
Who founded Sama and when?
Sama was founded in 2008 by Leila Janah as a nonprofit connecting workers in developing regions to fair-wage employment. The company restructured as a B Corporation in 2019 while maintaining its social mission. Janah passed away in January 2020, and Wendy Gonzalez became CEO.
What certifications and quality standards does Sama hold?
Sama holds ISO 9001, 27001, and 42001 certifications, plus GDPR compliance. The company maintains a 99% first-batch acceptance rate and employs approximately 4,000 full-time data experts—not crowdsourced workers. It achieved 8-year average customer retention with notable clients including NASA, Microsoft, and Sony.
What is Sama's Bulk Annotation feature?
Bulk Annotation is a 2025 feature that reduces labeling time by up to 80% and improves consistency by 25%. The tool streamlines data annotation workflows for large-scale AI training projects, enabling faster throughput without compromising quality standards or the company's 99% first-batch acceptance rate.
What happened with Sama and OpenAI's ChatGPT safety work?
In 2022–2023, reporting revealed Kenyan workers earning $1.32–$2 per hour labeled toxic content for OpenAI's ChatGPT safety training. Workers reported severe psychological trauma from exposure. Sama terminated the contract early in February 2022 under media pressure following publication of these allegations.
Why did Sama lay off 1,108 workers in Kenya in 2026?
In February 2026, Meta discovered Sama employees in Kenya had accessed private footage from Ray-Ban Meta AI glasses. Meta terminated the contract, triggering the layoff of 1,108 Nairobi workers in April 2026. This represents approximately 25% of Sama's workforce and underscores severe client concentration risk.
How Sama compares
Direct head-to-head against 3 competitors. Picked by 7wData.
Sama
- Positioning
- Provides human-verified data annotation, validation, and AI model evaluation services for machine learning algorithms.
- Customer segments
- Autonomous vehicle and ADAS companies training perception systems on image and video data
- Strengths
- Achieved 99% first-batch acceptance rate and holds ISO 9001, 27001, and 42001 certifications plus GDPR compliance
- Watch for
- Severe client concentration risk: Meta contract termination resulted in 1,108 layoffs in Kenya (April 2026), wiping out ~25% of reported workforce
- Recent moves
- Sama lays off 1,108 Kenyan workers after Meta terminates data annotation contract
Scale AI
- Positioning
- Full-service data annotation, RLHF, and model evaluation for generative AI labs, enterprise, and US defense agencies.
- Customer segments
- Generative AI labs, US defense and intelligence agencies, and enterprise automotive and robotics teams.
- Strengths
- Largest known RLHF and post-training annotation operation for frontier labs, with signed contracts from OpenAI and Nvidia.
- Watch for
- Meta's June 2025 acquisition of 49% stake and CEO Wang's departure to Meta raises strategic independence concerns among enterprise buyers.
- Recent moves
- Meta acquired 49% non-voting stake for $14.3B in June 2025. Founder Alexandr Wang replaced as CEO by Jason Droege.
Appen
- Positioning
- ASX-listed crowdsourced annotation and search relevance labeling vendor serving tech platforms needing broad-language training data coverage.
- Customer segments
- Large tech platforms and search providers requiring multilingual relevance and NLP annotation across 235+ languages and 170+ countries.
- Strengths
- Contributor network of 1M+ across 170+ countries, enabling multilingual dataset breadth no full-time workforce competitor replicates.
- Watch for
- Google contract termination in March 2024 removed $82.8M in annual revenue (approx. 30% of FY2023 total), client concentration remains unresolved.
- Recent moves
- February 2025 layoffs cut 48 employees and reduced LLM specialist team hours, despite that segment generating most 2024 profit.
Labelbox
- Positioning
- Annotation software platform plus managed expert network for enterprise ML teams operating in-house labeling pipelines with quality analytics.
- Customer segments
- Enterprise ML teams at tech, agriculture, and manufacturing firms managing multi-project annotation workflows with internal or contracted staff.
- Strengths
- Model-assisted labeling SDK with active learning and per-labeler quality analytics, measurably reducing manual annotation volume for in-house teams.
- Watch for
- Last institutional funding closed January 2022 at $110M total raised. No subsequent round in over four years signals capital runway uncertainty.
- Recent moves
- Acquired Upcraft, an agentic sales automation startup, in February 2026 to expand its Alignerr managed expert annotation network.
Sources
- www.sama.com — Company mission, products (Curate, Annotate, Validate, GenAI), customer testimonials, B Corp status, ISO certifications, 99% acceptance rate, 8-year retention claim
- www.vice.com — OpenAI contract details, worker wages ($1.32–$2/hour), toxic content exposure, psychological trauma reports, Sama early contract termination in February 2022
- www.theregister.com — OpenAI–Sama partnership, Kenyan workers classification of offensive ChatGPT training content, wage details, psychological impact documentation
- www.washingtonpost.com — Meta contract termination, 1,108 Nairobi layoff announcement (April 16, 2026), client concentration risk, transition support details
- thenextweb.com — Ray-Ban Meta AI glasses privacy scandal, Sama workers accessing intimate footage, contract termination trigger (February 2026), regulatory investigations
- www.sama.com — Series B funding details: $70M round, November 4, 2021, Caisse de dépôt et placement du Québec lead, Salesforce Ventures participation
- www.sama.com — Leila Janah founding story (2008), mission and philosophy, nonprofit to for-profit restructuring (2019), Janah's death (January 2020), Wendy Gonzalez succession as CEO