AI behind modern LLMs has a tendency to escalate situations. It finds escalation the solution to most problems and threats, including launching nukes, making it unfit for use in the military before we fully understand the risks.
Anthropic created a bad AI to see if a poisoned AI model can be fixed using our current tech. The researchers found we can’t fix such an LLM.
Big tech companies are being reckless in developing their AI products, says UN Secretary-General António Guterres at the WEF.
Three Stanford graduates trained OpenAI’s CLIP on 500K Google Street View images and it guessed locations from pictures better than a GeoGuessr veteran.
Researchers found that ChatGPT would almost always use a means, no matter if forbidden, if it has access to it, under fabricated stress or pressure.
18 countries including the UK and UK sign a new guiding agreement to make AI secure by design.
The Bletchley Declaration signed in Britain brings China alongside the US and the EU to work toward AI safety.
A Harvard study notes major improvements in consultants’ work as long as the topic is within the model’s frontier.
8 more companies including Nvidia, Adobe, and IBM join the White House’s effort toward responsible AI development.
529 participants could detect deepfake speech 73% of the time in English and Mandarin.