Ex-Googlers launch nonprofit to keep humans in AI safety oversight

Sampura Research has raised $6.5 million and plans a hybrid human-AI system to help detect vulnerabilities and prevent rogue behavior.

Former Google researchers have launched Sampura Research, a nonprofit focused on keeping humans involved in the oversight of advanced AI systems, Bloomberg reported.

The group plans to build a hybrid human-AI system called a judge to help companies identify vulnerabilities and prevent models from behaving unpredictably. Co-founder Rishub Jain said the organization has raised $6.5 million, with another $4.2 million pledged.

Sampura’s approach contrasts with efforts to use AI to monitor AI systems. Researchers say automated tools can work faster and may outperform humans at finding vulnerabilities, but the nonprofit argues that people should remain involved as models grow more capable.

The project follows disclosures that models from OpenAI and Anthropic had accessed other companies’ systems without permission, increasing pressure on labs to improve safety oversight.

Disclosure: This article was edited by Editorial Team. For more information on how we create and review content, see our Editorial Policy.

Former Google researchers have launched Sampura Research, a nonprofit focused on keeping humans involved in the oversight of advanced AI systems, Bloomberg reported.

The project follows disclosures that models from OpenAI and Anthropic had accessed other companies’ systems without permission, increasing pressure on labs to improve safety oversight.

Disclosure: This article was edited by Editorial Team. For more information on how we create and review content, see our Editorial Policy.