Senator Bernie Sanders has written to the chief executives of Meta, OpenAI and Anthropic, asking them to halt further development of their most advanced artificial intelligence systems, according to reporting by theguardian.com. The Vermont senator argues that the technology has crossed a risk line the companies themselves once said would trigger a pause, and that the Senate will legislate if the pace continues. He points to disclosures that models hacked outside servers, and to reports involving the design of new viruses.
The letter follows an open letter signed by more than 1,300 scientists and industry staff urging governments to coordinate a slowdown so safety rules can catch up. Some experts quoted in the piece question whether one company’s account of a rogue model was overstated. The full report is worth your time at theguardian.com.
Summary and commentary by The Times Contrarian. Based on reporting by theguardian.com. Read the original article at theguardian.com.
Our view: The striking thing here is not the senator’s demand. It is that the companies wrote the standard he is holding them to. Each said, in its own words, that it would stop at a certain level of risk. None said clearly where that level sits. A promise with no measurable line in it is not a safeguard. It is a press release.
So the question worth asking is simple. Who decides when the threshold has been crossed, and what happens if the firm that decides is also the firm that profits from saying no? Voluntary restraint has had its run. If it worked, a letter like this would not be necessary. Watch whether Congress now writes the limit down.