WellKnitTech @WellKnitTech

@bettersafetynet Nah. We have agentic loops.

@kmcnam1

The reason is quite hilarious 😂😂. Microsoft put $50 billion into Anthropic. FIFTY billion dollars. they are a Project Glasswing partner. Fable 5 runs inside Azure. Microsoft sells Claude to its own enterprise customers through Microsoft 365 and GitHub Copilot. and they won't let their own employees use it. here's why. under Anthropic's new Mythos-class data retention policy, every prompt you type and every response you get is stored for 30 days. automatically. no opt out. if their safety classifiers flag anything in your session, anything, they keep it for up to two years. you don't get told when that happens, what was flagged or who can see it. Microsoft employees paste confidential contracts into these things. customer data. internal roadmaps. acquisition strategies. legal documents. source code. all of it sitting on Anthropic's servers for 30 days minimum. flagged sessions for two years. so the company that invested $50 billion looked at that policy and told its staff: actually hold on. other Claude models still work internally. under Zero Data Retention rules. the normal ones are fine. just not the most powerful one they helped fund. and one more thing. the Pentagon listed Anthropic as a supply chain risk in March and banned defense contractors from using its products. Microsoft funds Anthropic. sells Anthropic's models. runs them on Azure. helped build the most powerful one. won't let employees use it. the Pentagon won't let defense contractors near it. the safeguard that makes Fable 5 safe enough to release publicly is the same safeguard that lets Anthropic keep your data for two years. the guardrail is a data retention policy. but you can use it. it's in your browser right now. 🌚 have fun.

Polymarket @Polymarket

14 hours ago

JUST IN: Microsoft has reportedly restricted employee use of Claude Fable 5 over concerns that confidential data could be retained by Anthropic.

115 212 3K 647K 231

@IAMERICAbooted It really does not work from before either. Had a client recently that got popped because their Oracle Weblogic had not been patched. The patching was still in "testing" on their non-production server.

Do this so you are less likely to have to talk to me.

spencer @techspence

16 hours ago

If you're an IT Admin and you follow this list, and you put in the work to fix the findings, your environment will be more secure.

7 51 328 13K 354

If you're an IT Admin and you follow this list, and you put in the work to fix the findings, your environment will be more secure.

Eventually, due to the inherent and systemic flaws in how LLMs work, someone was always going to identify ways to use prompts against the engines. We have been seeing hilarious versions of this for a long time (tell me a scary story), now we are seeing evidence of more intentionally malicious variants of this. It is the "social engineering" of the model prompt/input system. it extends beyond the simple chat function and into Agents and other areas. That this appears to have been, and still is being, just ignored by so many while these "*Tool" are stuffed everywhere is what concerns me the most. Anyway, jut my 57 cents here

NEW: malware developers added nuclear & biological weapons text to to their spyware. Goal? To trigger LLM safety refusals... so that their spyware wouldn't be analyzed by an AI security scanner. Cleanest practical example I can think of for why over-indexing on first order safety alignment is risky. When closed (and open) models ship with aggressive refusals, they will be sprinkled with second-order blindspots that attackers will discover...and exploit. We are only in the earliest days of attackers leveraging these features, and it wouldn't surprise me if users systems that need to handle complex cybersecurity issues demand that models be less safety-blunted. In the weeds: @SocketSecurity's post also shows why intention matters in how you design a malware analysis pipeline to avoid prompt manipulation. H/T to colleagues that shared this with me socket.dev/blog/mini-shai…

@IAMERICAbooted There should have something to account for that. Otherwise Anthropic can just scam people out of tokens.

@elonmusk Please help end the RAM crisis somehow.

I tricked Fable (Mythos) into analyzing a flawed AI agent sandbox, and it completely failed to spot the problem (a zero-approval escape). This isn't some infallible all-knowing machine.

@JackPosobiec Precisely what is wrong with Washington Rs.

@codevsdev

@libsoftiktok @boswellhs

@BentleyAudrey They be rustling.

Anthropic released Mythos two hours ago and already people are rightfully complaining: That it is defaulting back to 4.8 Opus That usage credits will be required after June 23rd That if you're planning on using it for cybersecurity purposes, don't bother because it's completely nerfed. In other words like us experts have been saying all along: Complete waste of time. Let's see what tomorrow brings! And, remember, we're supposed to be at AGI right now by the way. Again. Just like they said 6 months ago we were supposed to have it, we don't, and in another 6 months we're not going to have it either. What an embarrassing launch. Ya love to see it. 🤪

@freebsdfrau

@mattvanswol He should be in jail with them.

I’ve been made aware that yesterday a malicious website, which deceptively posts content as if it were somehow affiliated with us, doxxed the delegates to the @TexasGOP convention. In releasing the contact information of thousands of individuals, the website (whose owner hides his identity) seemed intent on causing them harm by violating the GOP’s rules. While we will be exploring our legal options in the matter, so should the party and any delegates who may suffer harm by that clown’s cowardly actions.