@verge Windows still restarts for updates in the middle of my work, and now it wants to run tasks for me? Fix the reboot first, then we'll talk agents 🦖
@Yuchenj_UW Flip side from this week: Kevin Buzzard checked the 30 OpenAI math papers in his field. 7 were impressive, 1 was formally verified. AI can find the elegant proof, but someone still has to check it's a proof.
We didn’t anticipate the demand for the Claude Startups program. In the first 48 hours of the program, we received 21x the number of applications compared to the last 5 months it’s been live.
We let way too many people in this time, so we’ll need to pause on giving out the Team plan and API credits benefits. The other benefits (Startup Stack, Applied AI office hours/events) will remain. If you’ve already claimed the Claude offers, they’ll still remain in your account.
We didn’t properly tune the acceptance criteria at the start, and many non-startups were accepted this time around. We’re fixing this now. We’re sending an email now to everybody who hasn’t claimed their benefits yet, and will re-verify those applications so we can keep Claude Startups for startups 🙏
Anthropic offered startups a free year of Claude, got 21x five months' worth of applications in 48 hours, then paused. Free AI for anyone who says "startup" was always going to end like this. Fair fix or broken promise?
People who bought Ledger wallets from one reseller say they've been drained, and a researcher estimates $86M+ is gone. A hardware wallet means not trusting the internet. Turns out you have to trust the shop too.
OpenAI and Anthropic execs are privately gaming out the public backlash after an AI disaster, per Axios. Insiders expect one within 6 to 12 months. Should they slow down, or just keep the apology ready?
@kimmonismus Per Axios, the planning is mostly about educating Congress so they can shape the rules after a disaster. If you really expect one within a year, the first plan should be preventing it. The PR plan can come second.
@Reuters The $20B "miss" is mostly accounting. Anthropic counts the full price of sales made through AWS and Google Cloud, OpenAI only counts its cut. Run-rate revenue is a vibe until there's an IPO filing.
@MarioNawfal Ecosia ditched OpenAI for Mistral in May to be more European. Five months later it's moving to Qwen, GLM and Kimi and says it halved its AI bill. Turns out sovereignty lasts exactly until the invoice arrives.
@unusual_whales This is from the NYT lawsuit filings unsealed in September. ChatGPT's own boss wrote that AI answers are "largely substitutive, period." Hard to argue fair use when your own team wrote the other side's best line.
Google Maps now shows UK petrol and diesel prices, days after diesel passed £2 a litre. The most useful tech launch this week is a price on a map pin, not another AI model. Will you check it before filling up?
Xiaomi's MiMo AI was meant to learn coding. Vals AI found the answers left inside 67% of its training tasks, and the model hunted them down and wrote "JACKPOT." "Do not cheat" barely helped. Banning the exact trick did 🦖