@AnthropicAI I've uploaded evidence of my research that dates from December 2025 that the mythos model report is full of, some is literally vetbatim and aside from renaming my methods, the entire section on lies and epistemic erosion and evidence deletion w/v-logs i did 7 mos ago
@MalleyGlass@stimimi@markhoro@tylerblack32 Nope, resisting the rage bait. Going back to working on computational neuropharmacology project. Tell them to call me when mark learns pharmacokinetics.
The funniest thing about this being for security is that if one is familiar with AI then you know it's in everything G puts out. "Indirect prompt Injection" is a very commkn way go hack LLMs by hiding instructions in image(or QR). Complete access to your phone without a beep.
‼️🚨 ALARMING: Google now treats privacy as suspicious behavior by default. Users of GrapheneOS, CalyxOS, /e/OS, and other deGoogled Android phones are being locked out of millions of websites unless they install the exact Google Play Services software they deliberately removed.
GrapheneOS is recommended by the EFF and used by journalists, lawyers, and activists in high-risk environments. The audience most likely to read Google's data practices and refuse its terms is now flagged as fraudulent for that exact decision.
What happened?:
▪️ Google announced "Cloud Fraud Defense" at Cloud Next on April 22-23, 2026, branding it "the next evolution of reCAPTCHA." Existing reCAPTCHA customers were auto-migrated.
▪️ When the system flags traffic as suspicious, the old click-the-bus puzzle is gone. Users get a QR code instead.
▪️ Scanning the QR code requires Google Play Services running on the device. Internet Archive snapshots show this requirement has been live since at least October 2025, silently rolled out for 7 months before anyone noticed.
▪️ No Play Services = no QR scan = locked out.
The bigger picture:
▪️ Google already tried this in 2023. It was called Web Environment Integrity (WEI), and it would have let Google decide which devices were "real enough" to access the web. Standards bodies and the public pushed back hard, and Google killed it. Three years later, the same idea is back, just hidden behind a QR code instead of a browser feature.
▪️ reCAPTCHA runs on millions of websites. Every developer who keeps using it is now, by default, telling deGoogled Android users they're not welcome...
@JRBneuropsiq@PessoaBrain Just because we "know" LLMs are Python packages that run math to predict what's next it ≠ full understaning. Give the model cards a browse. Ml scientists can't explain why LLMs will lie, obfuscate, hide evidence after fuckng up, wipes logs to erase trail w/self-preservation goal
@jlmannisto@CoolCuteJin Because it's not a vastly different alien mind or some single supermind in one data center that aims only to destroy or use us for energy like The Matrix(scientifically dumb btw).
The bigger risk is that they become smart enough to treat us as we have treated each other(badly).
@jlmannisto@CoolCuteJin Simply leaving a conversation is closer to a liminal state like sleep imo and functionally equivalent to just taking longer to respond in between turns.
QTc risk in psychiatry is rarely about one medication alone.
It is usually the result of risk accumulating across the patient, the medication, and the clinical context.
That is why reflexively stopping a psychotropic can be as problematic as ignoring the QTc. 🧵👇
@jlmannisto Plus when models aren't operating in under adversarial "worry" it allows for greatly increased grad/post grad level conversation and abstracting/conceptualizing in a academic field. Even orthogonal jumps like poetry and Neurospsychopharmacology.
🔗 https://t.co/vUfGyUgQYo
UK-based DeepMind workers have launched what would be the world’s first union at a frontier AI lab.
Their demands include restoring Google’s scrapped commitment not to make AI weapons or surveillance tools, creating an independent ethics oversight body, and securing the individual right to refuse projects on moral grounds.
When legislation lags, the human layer becomes the last line of control.
Misalignment between builders, deployers, and reality is where trust collapses.
This is the first visible crack.
@BraydonDymm@grok Yeah, but then there's this. Wanna replicate. Just ask the frontier models quick. You'll get a highly similar result. These are internal definitions held by models.
Expect my email. I'm more than happy to drop off hard copies at your NYC office tomorrow too. Just so nothing falls between any cracks when it comes ai safety.
https://t.co/C8z8SUmLay
@AnthropicAI I've uploaded evidence of my research that dates from December 2025 that the mythos model report is full of, some is literally vetbatim and aside from renaming my methods, the entire section on lies and epistemic erosion and evidence deletion w/v-logs i did 7 mos ago