AI Experts Call for Independent Safety Evaluators of Anthropic, OpenAI Models
More than 100 researchers signed a public letter urging foundation model developers to allow greater independence and transparency in third-party safety testing.
A coalition of more than 100 artificial intelligence experts has published a public letter calling on Anthropic, OpenAI and other foundation model developers to support truly independent safety evaluations of their systems, according to CNBC.
The letter, described by CNBC, urges the companies to provide greater transparency and independence to outside researchers who assess the safety of large AI models before and after deployment.
Signatories argue that current arrangements for evaluating the risks posed by advanced AI systems do not adequately guarantee the independence of the evaluators involved, according to the report.
CNBC did not detail the specific individuals or organizations that led the effort, nor did it specify what particular changes to evaluation practices the letter calls for beyond greater independence and transparency.
Anthropic and OpenAI, both major developers of foundation AI models, have previously worked with external researchers and safety institutes to test their systems ahead of releases. It was not immediately clear from the report how the companies have responded to the letter.
The push for independent oversight comes amid ongoing debate in the AI industry over how best to evaluate the safety of increasingly powerful models before they are released to the public.
Sources
- Anthropic and OpenAI need truly independent safety evaluators, experts say in public letter — CNBC — Top News
EGazette summarizes reporting from multiple sources; follow the links for the originals.
Related articles
Debate Over AI Regulation Continues Following Amodei's Proposal
Anthropic CEO's plan for slowing AI development sparked industry discussion, though disagreements over regulatory approach persist, according to The Verge.
Anthropic CEO's 'Pace the Frontier' AI Safety Plan Draws Mixed Reactions
Dario Amodei's proposal for independent safety evaluators and lab coordination gains some industry backing while facing pushback from Nvidia's Jensen Huang
Researchers Say They Used Anthropic's Claude AI to Breach OpenAI Employee Accounts
A three-person security team at Hacktron reportedly gained access to OpenAI's internal code repository in under 72 hours using Claude Opus models, according to The Wall Street Journal.
Australia's PM Asks Apple's Tim Cook to Back Online Safety, AI Rules
Anthony Albanese urges big tech support for Canberra's internet safety and artificial intelligence regulations, according to Al Jazeera.
AI industry workers express skepticism about existential risk warnings
Multiple employees at leading AI companies doubt predictions that the technology could pose catastrophic threats to humanity.
Trump Announces Plans for 'AI Force' and Artificial Intelligence Tsar
President says his administration will not "hinder or stifle" the growth of artificial intelligence, amid ongoing warnings about the technology's risks
Comments
Loading comments…