Anthropic co-founder Jack Clark reportedly warned recruits about a superhuman coding AI with nation-state hacking capabilities
He reportedly advised the recruits to pursue non-computing hobbies
Positive users find the Anthropic source credible and the pitch honest, while negative users mock the claims as baseless sci-fi marketing or blackmail.
Most Activity
Fake news because it’s from the Polymarket account
Truth because it’s Anthropic
@MatthewBerman Agreed. Why do I see this Polymarket account now all the time and its trying to pretend like its news?
@_arohan_ But that's what Jack and Dario have been saying since ... forever? (I also think it's not a particularly productive conversation, but really it's very par for the course)
This is really a silly thing to say. Love you guys but this is weird.
@MatthewBerman Lol they spend more money on marketing than compute; every day another baseless vague claim
@MatthewBerman Polymarket bets are just market sentiment; Anthropic's 'truth' comes with a very large compute bill.
@MatthewBerman And why would that stop someone from doing computer stuff for fun? Dumb take.
And if they have this superhuman coder they might wanna ask it to fix their up time and endless issues..
@MatthewBerman truth filters by how much u wanna believe it already
depressingly easy choice here
Anthropic’s framing turns containment quality into the market signal.
If a tool can write production code and lower the cost of intrusion, the buyer is not just buying throughput. It is buying sandboxing, audit trails, permissioning, kill switches, and liability allocation.
That shifts value toward vendors whose controls survive procurement and filings, not just benchmark screenshots. The downstream consequence is slower but stickier enterprise adoption: fewer casual rollouts, more governed seats, higher proof burden for renewals.
The clean check is filing durability: do customer 10-K risk factors, cyber insurance language, and vendor contracts start naming agentic code execution as a distinct control problem?
@MatthewBerman truth moves fast when the right account says it
@MatthewBerman Somehow Polymarket has actually become one of the most legit sources of news headlines so I kind of believe this one 😂
@MatthewBerman the anthropic source makes it way more believable than polymarket ever could
the nerd elite vs degenerate truth pipeline is real
@MatthewBerman I would be quicker to believe Polymarket than the company that says their AI models have a “soul” and LARP bad sci-fi movies
@MatthewBerman I like that he said hobby. Not job. The hope for UB/HI is still alive
@MatthewBerman meanwhile, Claude just deleted active directory to save space
@MatthewBerman A superhuman coder sounds cool until the security model is also superhumanly vague.
@MatthewBerman I like posting on X... new narrative stories like this on spacex bottlenecks (or lack there of)
@MatthewBerman It sounds like blackmail. Buy our stock so you can afford to do hobbies and live . America will not blackmail the world. We will isolate America shift to china
@MatthewBerman What a joke
@MatthewBerman Fake because Polymarket, true because Anthropic. I buy both. 'Get hobbies' while you're building your replacement. Most honest pitch in AI.