About this Concept
Safety & governance
Ways to reduce harm, set limits, test behavior, and govern how AI is used.
Trend this week
of conversation
9 of 9 weeks tracked had mentions across multiple podcasts.
11quoted moments6episodes6podcasts discussing this topic
A closer read
- What changed
- Its share fell from 11.1% to 10.7%, with evidence across 9 podcasts.
- Where it showed up
- It appeared in episodes including “Stealing Reasoning Traces from Proprietary LLM APIs — Ilia Shumailov & Al...” and “AI in the AM — Weekly Highlights: Relaunch Week (Aug 17–20, 2026)”.
- How to read it
- 6 cited episodes from 6 podcasts are shown below. The topic appeared across 9 podcasts in the full analysis.
Podcast moments from the week
Moments are grouped by episode so repeated excerpts from one conversation do not look like separate sources.
Machine Learning Street Talk
Stealing Reasoning Traces from Proprietary LLM APIs — Ilia Shumailov & Alexander Panfilov
“I see. Wow. It's kind of cool, isn't it? Okay. So, and can you talk through, presumably outside of privacy, there are also safety implications around this?”
ConceptSafety & governance
Report this moment“Who can you harm the most with all of the data that you've extracted?”
ConceptSafety & governance
Report this moment“let's be very cold-headed scientists and say, build controlled environments and make very precise assessments”
ConceptSafety & governance
Report this moment
The Cognitive Revolution
AI in the AM — Weekly Highlights: Relaunch Week (Aug 17–20, 2026)
“Then the harms map: cyber, bio. And the risk, he says, can't be clawed back.”
ConceptSafety & governance
Report this moment“where somehow it seems like no model is ever high risk according to internal evals. It's either low or medium. And that just is suspicious”
ConceptSafety & governance
Report this moment“I'll start with the obvious. Agent-orchestrated attacks are real. This wasn't intended to be a demonstration of AI cyberattacks, but we have that one. Threat actors that are intentionally optimising models and creating harnesses for offensive purposes can probably do a lot wor...”
ConceptSafety & governance
Report this moment
Latent Space
Simulation: the new Scaling Law — Joon Sung Park, Simile AI
“we also want to make sure that we form enough of guardrail and perspective on how to leverage this technology before we go on to serve markets like the politics”
ConceptSafety & governance
Report this moment
The AI Daily Brief
9 AI Techniques You Probably Haven't Tried
“this is a change that could help make AI in the enterprise less of a trade-off around risk and security.”
ConceptSafety & governance
Report this moment
AI Today Podcast
Critical Thinking and Trust in AI Projects: Insights from Cheryl Howard
“we don't want people to use it outside of the guardrails. And that's where governance is such a big deal to me.”
ConceptSafety & governance
Report this moment
The Artificial Intelligence Show
#232: Claude Watermarking, AI’s Environmental Impact, OpenAI Talent Drama & Anthropic’s Hidden Advisor
“OpenAI employees have discussed providing the detector directly to educators. or to outside companies that help schools identify AI-written papers and plagiarized work. Google has developed a watermarking tool that can detect text generated by Gemini AI called Syth-ID. It is i...”
ConceptSafety & governance
Report this moment“This was an idea he first made public in mid-July, a US-led standards group modeled on the FINRA financial services regulatory body that would safety test frontier AI models for dangerous capabilities before release. Apparently, Demis discussed the proposal privately with Trum...”
ConceptSafety & governance
Report this moment