From CBS News · · 1m 10s
Amodei wants AI rivals checked by outsiders
Anthropic's CEO says the companies building AI shouldn't be the only ones grading its safety.
In brief
Anthropic's CEO says the companies building AI shouldn't be the only ones grading its safety. Amodei is pushing the AI industry toward outside safety review instead of self-assessment. Originally reported by CBS News.
A call to the competition
Anthropic CEO Dario Amodei is asking rival AI companies to agree on shared safety measures rather than each setting its own bar.
Bring in third-party evaluators
His specific ask: let outside evaluators test AI systems. The people building the models would no longer be the only ones judging them.
“third-party evaluators”
Safety as a group project
Amodei framed safeguards as something the industry should pursue together, not a race where caution costs you the lead.
“encouraged the artificial intelligence industry to come together in the pursuit of safeguards”
Said on the record
He made the case in an interview with CBS News, taking an internal industry debate to a mainstream audience.
Airing September 13
The full interview runs on CBS Sunday Morning on Sept. 13.
Why the ask matters
An agreement only works if competitors sign on. One company inviting scrutiny changes little; all of them doing it changes the standard.
What matters most
Amodei is pushing the AI industry toward outside safety review instead of self-assessment.
