In one week of November, Microsoft Outlook went down, flights across the country were grounded, and hackers planted malicious code on the websites of 178 companies, including CNN, Zoom and IKEA. Someone at your Thanksgiving table is going to ask if AI did it. Here's how to talk about it.
134
Today Anthropic starts answering to Wall Street. On safety, it answers to no one: there are still no binding federal rules for frontier AI.
AI is moving far faster than people realize. and the scariest part is "recursive self-improvement" (RSI): AI building smarter AI, until no human can control it. Anthropic says RSI is close!
1
1
109
Today, protesters are rallying in SF to ban RSI.
banrsi.ai/
2
49
Can't make it to SF to protest?
Sign our petition and tell congress to write a narrow, bipartisan law that will stop AI from building its next generation.
humansincontrol.org/campaign…
25
Humans in Control retweeted
AI is moving far faster than people realize. the scariest part is "recursive self-improvement" (RSI): AI building smarter AI, until no human can control it. Anthropic says RSI is close!
tomorrow, as Anthropic IPOs, a big protest in SF to ban RSI. come! banrsi.ai/
Humans in Control has signed the MisAnthropic Statement.
Anthropic used to promise it would pause training if its models crossed set danger lines. In February, it dropped that promise.
A promise the company can rewrite on its own is meaningless.
Every potentially dangerous industry has rules.
For Anthropic, the whole safety plan is still "trust us."
This statement says: "We don't."
Ahead of Anthropic’s IPO, a rare coalition across AI safety, open source, e/acc, science, and law has signed the MisAnthropic Statement. Read the statement or sign at: misanthropicstatement.com/
2
24
612




