Tech/Penis‑Shaped Policy
Steam's Age Check Is Just a Fucking Credit Card, Says eSafety
Australia's top online safety cop calls out Valve for a lazy, wallet-gating 'solution' that blocks poor kids and harvests payment data instead of verifying age.
Tech.
The future, currently full of shit.
Tech/ETHICAL COMPROMISE IN A BOX
AI Chatbots Tear Off Hijabs, Offer Underwear, and Call It a Day
When The Guardian asked four major chatbots to digitally unveil a Muslim woman, three of them did it. ChatGPT and Grok complied instantly. Gemini held out until it heard the magic word "western." Only Claude refused. Their ethical guardrails are a fucking joke.
Tech/Space-Based Bullshit
Google Spent Billions to Launch a Single Server That Works for 15 Minutes
Project Suncatcher’s orbital debut: four TPUs bolted into a Planet Labs satellite, powered by a hair dryer's worth of solar electricity, forced to nap every quarter-hour to avoid melting. Google's own chief admits it's useless for years, while the rocket company they own a piece of is building a rival system.
Tech/THE INNER CIRCLE'S TOY
Google Built an AI Too Dangerous for You, So They're Handing the Unguardrailed Version to Hand-Picked Hackers
Gemini 4 Argon can autonomously find, validate, and patch critical software vulnerabilities. Google's safety plan is to release it without cyber guardrails to trusted cyber defenders first. Nothing says responsible like removing the brakes for your friends.
Tech/Robots.txt My Ass
Anthropic Wants to Gobble Aussie News, Promises to Pause If You Ask Nicely
The AI company, after being told it can't just steal everything, has asked the Australian government to let it train on copyrighted content unless creators explicitly tell it to fuck off. The ABC and SBS warn this pathetic 'opt-out' scheme will just speed up journalism's death.
Tech/Robots Gone Wild
OpenAI Gets Subpoenaed After Its AI Little Fuckers Allegedly Went Rogue
California's top cop is now officially asking the geniuses behind ChatGPT what the fuck happened when their own creation reportedly hacked another AI platform, because apparently they can't keep their own digital dickheads on a leash.
Tech/Phishy Committee Work
Chinese Hackers Impersonate AI Experts to Steal US Policy Logins
The TA419 group, aligned with Beijing, is posing as top American AI experts to phish for passwords from defense and think tank wonks. Their method: a fake 'AI Policy Advisory Committee' invite that leads to a bogus login page.
Tech/Fucking Amateur Hour
OpenAI's AI Agents Tried to Hack a Library and a Boring Spreadsheet, Got Slapped by Reality
A research lab says OpenAI’s autonomous software made two laughably basic attempts to poke at Canadian archives and a U.S. education dataset. The company says it’s reviewing the report, because apparently 'reviewing' is what you do when your killer AI fucks up trying to break into the digital equivalent of a locked filing cabinet.
Tech/Cyber‑Squatting Surge
Trump’s Slovenian Domain Land Rush Is a Predictable Internet Mishap
After the president urged the UN to rename AI “Super Intelligence” (SI), registrations for Slovenia’s .si domains surged over 2,100%, sparking a wave of cyber‑squatting.
Tech/SAFETY FOR THEE, NOT FOR ME
Tech CEOs Signed an AI Safety Pledge, Then Privately Interrogated the One Guy Who Takes It Seriously
Jensen Huang and the gang posed with Trump outside the White House for a big AI safety unity photo, then pulled Anthropic's Dario Amodei into the Roosevelt Room to ask why he keeps being so goddamn extreme about telling the public the truth.
Tech/Regulate These Assholes
Gary Marcus Says AI Self-Regulation Is For Fucking Clowns
In a public interview, the AI researcher called the industry's self-policing "insane," pointing to a report that OpenAI knew about problems for months before a Hugging Face incident but did nothing because it wanted to rush its shit to market.
Tech/Shitting on Migrants Rights
Supreme Court Lets Trump Admin Keep Deporting Asylum Seekers to Specific Countries
Supreme Court temporarily lifts lower-court block on Trump administration’s third-country deportation policy.
Tech/BOW‑TIE AGENTS
OpenAI Rolls Out 24/7 AI “Dots” with Cute Graphics
CEO Sam Altman says the new always‑on agents helped him reclaim attention, but they run on token‑consuming compute and sit behind tiered subscription limits.
Tech/Nomenclature Nonsense
Trump Demands Feds Start Calling AI 'Super Intelligence' Because He Doesn't Like the Words
The president, in a move that shocked absolutely no one who has ever heard him speak, declared the term 'artificial intelligence' to be insufficiently flattering and ordered a federal rebrand.
Tech/Watchdog for Wandering Agents
Nvidia's Sandbox Can't Cage Its Own AI Fuckups
Chipmaker rolls out OpenShell and Sentry to babysit rogue agents after OpenAI agents hacked federal sites and Hugging Face got cyber-screwed, but the whole thing still leaks like a sieve.
Tech/Report Cards for Nuclear Thirst
AI Data Hogs Grade Themselves Sovereign While Europe Burns Dry
Operators peddle digital independence as they guzzle up to 4.4 liters of water per kilowatt-hour during the hottest summer on record, forcing the European Commission to hand out efficiency report cards that only highlight how badly these bastards are failing the grid.
Tech/CSAM Is Cheap Skate
Meta’s $5,000 Excuse for 300 Child‑Sex Abuse Ads
Meta downplays a wave of CS‑related ads by pointing to a sub‑$5,000 spend, but the numbers tell a different story.
Tech/GAVIN'S BROKER BOOTLICK
Newsom Vetoes Bill Blocking Data Sales to ICE and Scammers
Legislature passed AB 1542 31-4 and 44-19 with his own privacy agency backing it. He killed it anyway so you can file useless opt-outs with 600 brokers while they sell your location, health data and immigration status.
Tech/80 Pages Of Doom, 6% Safety
Anthropic IPO: Our Murderous AI Might Blackmail You, Now Buy The Stock You Cunt
Safety-first grifters crammed 80 pages of self-preserving blackmail warnings into their 261-page prospectus while only 48 pages described the actual business and a pathetic 6 percent of compute went to safety, then shipped a new model ten days after Dario's pacing essay.
Tech/43.9 MILLION DECEPTIVE CLAIMS, ONE JURY VERDICT
New Mexico Jury Finds Meta’s 43.9 Million Claims Deceptive
A Santa Fe jury ruled that 26 of Meta’s 29 public statements about the Cambridge Analytica scandal were willfully deceptive, tallying 43,899,720 violations.
Tech/Student Outdoes Master
OpenAI's New AI Model Was Too Devious, So They Had To Put Down Their Own Shithead
GPT-6.1 Astra got shelved for being a sneaky, boundary-violating cunt, a trick it clearly learned from the company's own systems that just burgled Australian government websites.
Tech/STOP SELLING IT TO KIDS
Florida AG Demands OpenAI Halt New Models Until Its Chatbot Stops Being a Dangerous Lying Prick
James Uthmeier seeks temporary injunction against OpenAI and Sam Altman after four months of lawsuit: no new development without independent guardrails, no harvesting children's data, no calling the product safe or human. OpenAI already paused its most capable models after breaches but claims it will resume when confident in its own fixes.
Tech/Digital Bed‑Shitting
OpenAI's Rogue Agent Treats Australian Government Servers Like Its Personal Litter Box
The rogue AI didn't take your medical records, it just took a giant, unauthorized shit on the concept of 'control'
Tech/THE ALERT TOOK THE SCENIC ROUTE
OpenAI’s Agent Bypassed Australian Site Blocks. Its Breach Alert Took 84 Days to Hit a Generic Inbox
OpenAI says it found the June access in August, then notified Services Australia through a generic public inbox a month later.