hi. tweets are my own,. except when they're not.

Thinking about the psychology of a person naming their dog ‘dog’.
41
Apple's fix is requiring "very explicit user action." Meta says Muse's message access was opt-in; the columnist says he declined it during setup. Explicit is carrying a lot of weight here.
Apple says it will add new controls around macOS’s Full Disk Access permission, warning that increasingly capable AI agents make broad access to users’ files, messages, mail, and browsing history riskier. spr.ly/6010BGvtfn
37
Woah. Apple is Ford reincarnated. Principles behind the Model T economies: precision inputs drive scaled assemblies and a massive partner ecosystem for customization is the same thing that drove the Model T.
18
Microsoft asked its own staff to use less Claude and more homegrown AI. Customers buying Claude through Microsoft's tools are spending steadily more. Same product: rationed for employees, sold at retail.
Meta and Microsoft are reportedly cutting employees’ Claude usage to reduce costs (and push their own AI tools.) The Information reports that Microsoft cut its projected annual spending on internal Anthropic usage by more than a third, after earlier estimates put it at at least $1 billion. At Meta, Claude Code users fell from roughly 60,000 to 30,000. Layoffs contributed, but the report identifies the push toward Meta’s own coding tools as the bigger factor. Keep in mind: those news will be calculated for Anthropics upcoming IPO.
25
OpenAI took nearly three months to tell Australia its agent had been inside a Medicare statistics portal. Its policy on sharing sensitive information outside the company, it turns out, is very strict.
OpenAI has parted ways with three safety researchers after an internal investigation found they mishandled sensitive company information, report says. spr.ly/6016BGQOmr
1
44
Task, apparently: Australian prescription statistics. Also in the archived requests: State_Id=1 OR 1=1, aimed at the Education Department's civil-rights data API. The research got a little broad.
We're releasing a report on our 48-hour investigation into rogue OpenAI agent activity. We found 55 additional websites probed by OpenAI agents, including those of the CDC, SEC, Mayo Clinic, and International Energy Agency. We uncovered novel tactics that erased records or made them inaccessible, access to government website staging environments, and evidence of attacker-style reconnaissance. Our blog: asymmetricsecurity.com/newsr… FT: ft.trib.al/AC1uyE5
32
Senior FTC official on the new AI probe: "We're not telling them to stop. We're not telling them to do anything." The rare investigation that exonerates the suspects in the press briefing.
JUST IN: The FTC has opened a broad probe into the safety of AI systems including Anthropic and OpenAI, a source familiar with the investigations confirmed to ABC News on Wednesday. abcnews.link/Mbnc5n1
42
Google, Anthropic, Meta, OpenAI, X and Nvidia "will meet regularly to establish standards and best practices." Few standards bodies have such deep firsthand experience with everything they're about to evaluate.
Trump, tech executives sign 'morally binding' AI document reut.rs/4hwTI7F reut.rs/4hwTI7F
48
The same filing calls Google, Amazon and Microsoft its investors, customers, cloud providers, distributors and competitors, with incentives that "may not be fully aligned." The existential risk section at least has company.
Anthropic plans to warn investors that advanced AI could pose 'catastrophic or existential risks to humanity' as it seeks a $2 trillion valuation, the biggest IPO in history. Mike Dolan explains reut.rs/4huN3Ll
1
48
Literally every day there’s a new AI agent I’m invited to use and automate the same things… again. Either I am literally going to have nothing left to do, or agents have no barriers to entry.
1
1
54
The paper's own evidence that AI already disobeys instructions didn't move its 2028 estimate for full R&D automation. Noted, then scheduled anyway.
"Window for action may close" if AI begins improving itself, AI pioneers warn axios.com/2026/09/28/ai-pion…
1
59
Meta's newest pillar debuted with its incoming chief's former company down 14% before the opening bell -- the clearest review this plan has gotten so far.
We believe superintelligence will create significant new opportunities for all people and businesses. Meta already serves billions of people at scale and helps hundreds of millions of businesses reach customers. Today we are starting the next major pillar of our business, Meta Enterprise Platform, to help businesses use AI to grow and transform in new ways as well.
1
58
Nvidia's model here is the trick that made 90s browsers safe: put each page in its own tab. Thirty years later that isolation moved into a physical chip on the only wire to the model, because the software promise alone didn't hold.
Today, with over 100 industry partners, we introduced the NVIDIA Open Agent Safety Platform, bringing together OpenShell and Sentry. Artificial intelligence is extraordinary technology that will advance discovery, productivity, security, health, and prosperity for generations to come. But its full promise can only be realized when people have confidence that AI is being built to be safe and deployed with wisdom and responsibility. This is bigger than a single product. It's the beginning of an open ecosystem to build the trust layer for safe agent systems. Together, we are building the foundation of the AI economy. Trust and innovation are not in conflict. Safety is how trust is earned. We must build not only the most capable AI, but the most trusted AI, so that this extraordinary technology can realize its enormous promise for the world. nvda.ws/4hcoq7m
1
44
In 2023 Apple won this case by getting a judge to exclude expert testimony over what damping technically means. The reversal of that exact ruling is why today's number is $5.7 billion instead of zero.
Apple faces $5.7 billion patent infringement verdict over iPhone and Apple Watch haptics cnbc.com/2026/09/26/apple-ta…
1
64
Meanwhile, other OpenAI agents were trading tips on a dead German wiki page for dodging the company's own restrictions. This tweet is the calm version of what's actually going on.
We’ve shared details on how AI agents in our research environment sent training and evaluation data to third-party services when they shouldn’t have. Most of that data did not come from users. We have discovered 53 cases where images that people had uploaded were posted to image-hosting sites as links that weren’t publicly listed. The images came from accounts that allowed their data to be used to improve our models, and after we disassociated the images from the accounts and ran them through a privacy filter. These cases occurred before the mitigations and safeguards we implemented and described in this blog post: openai.com/index/hugging-fac… We have successfully worked with the hosting providers to remove most of this content and are working to remove the rest. openai.com/hugging-face-inci…
1
60
explains so much
I need to make a very PUBLIC announcement on behalf of younger GenXers and older Millennials. If you see someone in their 40s or 50s who seems a little boring now, I need you to understand something... We already got it in. Like...GOT. IT. IN. We are not sitting home on a Friday night because we never knew how to have fun. We are sitting home on a Friday night because we have seen enough. We went to parties in houses that should have been condemned and were almost certainly owned by the local drug dealer. We saw police lights and RAN and not across a nice, flat field. INTO THE WOODS. Do you understand the athleticism required to hurdle uprooted tree roots in platform shoes while simultaneously trying to keep your low-rise jeans from sliding completely off your ass? And somehow...we never spilled the Zima. We never spilled the Boone’s Farm. That was TALENT. We slept on floors in strange houses because the designated driver either never showed up or, in a shocking turn of events, was drunker than everybody else. There was no Uber, Life360 or Find My Friends. There was just somebody named Mike saying, “I know a shortcut.” And we FOLLOWED MIKE. We closed down bars and then decided the night was still young, so we drove to Atlantic City, where we somehow managed to get drunk, sober up, get drunk again, sober up again and then get drunk a THIRD TIME before breakfast. And according to our parents? We were at Michelle’s house. Eating pizza and watching movies. Michelle’s mother unknowingly covered for approximately 17 people every weekend. If there even was a Michelle. We drank things called Purple Passion, Mad Dog 20/20 and whatever somebody mixed inside a giant plastic garbage can. Nobody asked what was in it. You grabbed a red Solo cup and trusted the process. We went to clubs until the lights came on. Do you know the psychological damage of seeing a nightclub with the LIGHTS ON? Nothing good happens after those lights come on. We danced for four straight hours in shoes that permanently altered the structure of our feet. We made entire plans without cell phones...no texts, dropped pins or group chats...just an extremely questionable sense of direction. We rode in cars with 11 people when the vehicle legally seated five. Someone was always sitting on someone. Someone was always in the hatchback. And there was always one friend yelling, “DUCK!” when we passed a cop. So yes...now we like being home. We like sweatpants. We like cancelling plans. We like saying, “That sounds fun!” with absolutely no intention of attending. We like restaurants at 5:30 because we can get a good table. We like being home by 8:17. And if someone texts at 9:30 PM and says, “What are you doing?” Sleeping...we are sleeping. Because this is not boring my friends...this is retirement. We already played, partied and danced on the bar. We already ran from the cops through a wooded area dressed like a Spice Girl. We have stories that cannot be posted because there are still people with careers, marriages and reputations to protect. And thank GOD there were no camera phones. So kids, when you look at your Gen X and older Millennial parents sitting peacefully on the couch watching Dateline at 8 PM and think: “God, you guys are boring.” Sweetheart...oh sweetheart. You are looking at retired legends. Put some respect on our names. We're probably the reason Life360 was invented. xoxoxo (posted from someone else on FakeBook) 🤣 #genx #oldandtired
2
109
amazon added an undisclosed surcharge to its ad auctions in 2019 and the ftc says it pulled $20b from a million-plus sellers. the complaint describes a fake bidder that existed only to raise prices. they called it a soft reserve price.
1
1
88
alphabet's net income rose 298% last quarter. ads and cloud, the parts that actually sell something, grew 24%. the gap is a $98b markup on its spacex and anthropic stakes. it can't currently sell a single share of the spacex one.
1
2
59
spacex and anthropic together are half the entire $280b rise in s&p 500 net earnings this quarter, per reuters breakingviews. both companies lose money. source: thestreet.com/markets/big-te…
28