The Hallucination HQ

Archives
Log in
Subscribe
1 September 2026

Today's AI News from The Hallucination HQ

The Hallucination HQ

AI news that's actually fun to read — 2026-09-01

Today's Hallucination HQ

The Pentagon Has Logged On

The US Department of Defense has launched a central AI portal giving military personnel access to ChatGPT, Grok, and Google's Gemini — a sort of very serious, extremely classified app store. The Pentagon joins a growing list of institutions discovering that the fastest way to modernise is to give everyone the same tools civilians have been arguing about on social media for three years. No word yet on whether the chatbots have security clearance.

Source: TechCrunch


Brussels Would Like a Word, OpenAI

The EU has classified ChatGPT as a Very Large Online Search Engine under the Digital Services Act — a legal label that sounds made-up but carries very real teeth. OpenAI must now actively mitigate ChatGPT's risks to minors, user mental health, and illegal content spread across EU member states. Think of it as the Brussels version of a school note home, except the school has regulatory enforcement powers and a fondness for fines with many zeroes.

Source: The Verge


Maths Is Fine. Probably. Don't Worry About It.

In May 2026, an OpenAI model produced a counterexample to the Erdős unit distance conjecture — a famous unsolved problem in mathematics — and five human mathematicians verified it the same day. A new academic paper now asks what happens to mathematical knowledge when AI can generate proofs faster than humans can check them. The short answer: verification is becoming cheap, trust is becoming expensive, and mathematicians are having a perfectly normal one.

Source: ArXiv


Nvidia Drops $3.5 Billion, Remains Relaxed About It

Nvidia has invested $3.5 billion into Taiwanese chipmaker MediaTek — a modest hedge against the growing threat of Big Tech building its own AI chips and quietly cutting Nvidia out. As Google, Amazon, and Microsoft develop in-house silicon, Nvidia's strategy appears to be: partner with the companies who supply components to everyone, including the rivals. It's less "staying ahead of the competition" and more "becoming structurally unavoidable."

Source: TechCrunch


Instagram Discovers That Fake People Should Be Labelled

Instagram is adding an "AI-generated profile" label to accounts that are, in fact, not people — including AI influencers like Aitana Lopez, a computer-generated model with a fanbase that would embarrass some actual humans. The platform is also renaming the existing "AI creator" tag, apparently concluding that clarity is preferable to the current system, which is no system at all. Better late than never, though "never" was looking genuinely competitive for a while.

Source: The Verge


As always, we'll be back tomorrow — assuming we're real.


**

Was this forwarded to you? Subscribe here

Written with AI assistance, edited by humans. • Privacy • Unsubscribe

Don't miss what's next. Subscribe to The Hallucination HQ:
← Newer The model that escaped Older → Every pharmacist is a hallway
Powered by Buttondown, the easiest way to start and grow your newsletter.