TechCrunch Host

speaker
5,755 appearances 109 recordings 1 series first heard Mar 2026 last heard 17 Aug

TechCrunch Host’s voice in public audio — every appearance, attributed to the second.

Trend

recordings per month · last 12 months
23 · Jul OctJan 26AprJulnow

Recordings per month over the last 12 months — 109 in all, peaking in Jul 2026 with 23.

Appearances

newest first · ▶ plays the moment
Ceylon said proper safety evaluations go beyond controls and containment of the environment. matched
There needs to be much better monitoring of the tests once they are underway. matched
She said, I think the interesting thing in several of these cases is that no one caught it when it happened. matched
OpenAI found out because of Hugging Face, Anthropic didn't catch it until they went back and looked. matched
Meta was similar. matched
I'm sure there were signals they could have detected. matched
In anthropic's postmortem of these three incidents, matched
The company admitted that both it and Irregular could have done a better job at monitoring, and that in some cases there were clear signs that something was amiss. matched
Experts also called for independent third-party audits of evaluation environments before models are unleashed in them. matched
Yoon said, if say Irregular had hired or been compelled to hire an external auditor to check the matched
configurations of their systems before running evaluations on them, they certainly would have caught the issue there. matched
Even if people had a meeting ahead of time to just go through the checklist, they would have called it. matched
Yeah, the fact that they did matched
shows that there's some very severe corner cutting happening. matched
A source familiar with the details told TechCrunch that irregular environments are continuously reviewed and tested, including in consultation with multiple external parties. matched
The source also said that monitoring was in place, but that monitoring isn't sufficient on its own. matched
Yoon and other researchers urged the industry to come up with a standardized process for frontier model safety evaluations. matched
He said, especially when the guardrails are turned off, you have to treat it like you're putting the most capable hacker in the world inside that environment. matched
The problem isn't that companies don't know how to build more secure testing environments, both Yoon and Bitterman argue. matched
It's that doing so can be expensive and cumbersome. matched
Showing 301–320 of 5,755 · page 16 of 288 ← Previous Next →