- About two hundred people marched past the San Francisco offices of OpenAI, Anthropic, and Google DeepMind, asking each company to promise to stop training frontier models if the others promise the same (Mitch)
- The author sees the demand as low cost for the companies and a possible push toward the international coordination that stopping the race would actually require (Mitch)
- Political campaigns are using chatbots to answer replies to canvassing texts, sometimes for hours at a time, and one vendor reports tens of thousands of conversations from half a million outgoing messages this year (Mitch)
- Disclosure laws exist in North Dakota and California, but the author doubts voters will reliably know when they are talking to a bot or who paid for it (Mitch)
- Reviewing AI 2040: Plan A, the author argues that even relatively lucky futures force hard choices about surveillance, weapons, employment, and corporate power, and asks critics what they would do instead (Mitch)
<- Back to digests
Daily digest
Marchers, campaign bots, and the case for having a plan
By Mitchell Howe
This issue covers a San Francisco protest calling on AI companies to stop racing, the spread of campaign chatbots that text voters at scale, and an argument that critics of the new AI 2040 scenarios owe the public an alternative plan.
Read the full dispatch on AI StopWatch
From AI StopWatch, published with their permission.
The summaries and the full Spanish translation are produced automatically. The original is always linked.