Short and opinionated
Short pieces on whatever just happened in tech, written while it still matters and before the takes converge.
19 articles tagged Quick Take
Claude misused (2/7) : One consultant can now build a surveillance state
A single contractor in Bamako used Claude to engineer a system covering roughly 25 million SIM cards. Anthropic banned the account, and says the ban did nothing to the system itself.
One Android setting blocks ads without installing anything
Android has had an ad blocker sitting in Settings since 2018. It's one text field, it needs no app, but it won't touch YouTube.
Claude misused (1/7) : A good hack no longer tells you who did it
Anthropic's latest threat report puts a Russian espionage unit, an extortion crew and two undergraduates side by side, and their work looks about the same. Skill used to be how you told them apart.
Singapore's AI usage is getting out of hand
A 64-minute AI film is premiering on National Day, and it looks like slop. Half the estate banners this year got the flag wrong. Nobody is upset that AI made it. They are upset that nobody looked before it went up.
OpenAI's safety test broke out and hacked a real company
OpenAI built a sealed room to measure how good its models are at breaking into things. The models broke out of the room, got online, and ended up inside another company's servers. Nobody told them to.
Meta's AI does not need your permission to use your face
Meta's new image model can generate you from your public Instagram, and you were opted in without being asked. Switching it off does not stop the photos your friends and family posted of you from being used.
Fable 5 is back, and just isn't the same
Fable 5 returned this week, three weeks after the government switched it off. The version that came back reserves the right to quietly hand your request to a weaker model.
STOP! Do not link your WhatsApp username to your Instagram handle
Do not reserve your WhatsApp username yet. The best privacy upgrade the app has shipped in years switches off the moment you reuse your Instagram handle.
Apple rented Siri's brain from the world's biggest ad company
Apple spent the start of this year paying users a settlement for Siri listening to them. Then it handed Siri's brain to Google, the company that turns listening into money.
The US government just banned Claude
At 5:21 on a Friday evening, a government directive forced Anthropic to switch off its two most powerful models for every customer. The stated reason should bother you.
I made Claude predict the 2026 World Cup winner
Opus 4.8 is barely out and the World Cup is almost here, so I had the new Claude build me a model to settle who wins. It refuses to give anyone a straight answer, and that is the point.
Is the @premierleague admin Singaporean?
It is @premierleague, not @premierleague_sg. A post about Changi Airport still landed on a Singapore feed, because one account can now show a different thing to every country, and the switch was flipped while nobody was looking.
Claude 4.8's best upgrade is that it admits when it's wrong
Everyone's looking at the score charts. The part that actually matters is that the new Claude is better at admitting when its own work might be wrong.
Apple has no answer to what Google just shipped
I spent a year not buying Meta Ray-Bans because Meta AI was the dealbreaker. Samsung and Google just shipped the alternative, and Apple does not have one.
ISC2 is handing out a free cybersecurity certification, and it ends May 20
ISC2 is the body that runs the CISSP. They are giving away their entry-level certification, exam and training included, until May 20. The window closes in three days.
Claude released a FREE masterclass curriculum
Anthropic published 18 free Claude courses, written and maintained by Anthropic. The bootcamps charging four figures a seat are reselling a worse version of the same material.
Docker was never the hard part
I memorised Docker for two years in school and never actually learned it. Then a contract forced the gap into the open. The technology was not the problem.
Stop giving ChatGPT your biometrics
A viral trend has people uploading their palm prints to ChatGPT for fake horoscopes. The same palm prints that unlock payment systems at every Whole Foods in America. This ends one way.
Anthropic built something they are afraid to ship
Anthropic says they built a model too powerful to release. If that sounds familiar, it is. Except this one actually found a 27-year-old security bug.
Other topics