Will Luttrell

Everyone's talking about what AI Agents can do. Not enough people are talking about what's stopping companies from actually deploying them. 🧵1/7

@honeycakeapp was built to solve this. Not only would the hacked files be un-openable, but we could even help track down the bad guys if they tried to open them. @LAPDChiefBeck @lapd

@chatgpt21 We believe that things like this will bolster the need for self-protecting files like we build @honeycakeapp. File location as a proxy for security will no longer be sufficient. Easy encryption, audit, and control will rise in importance.

I think we just got a demo of Mythos and I’m surprised nobodies talking about it.. 💔 In what might be the first instance the general public has seen of Claude Mythos Mythos (TBD) just uncovered a critical zero-day vulnerability in Ghost, an open-source platform with over 50,000 stars on GitHub that has never had a critical security flaw in its entire history. It identified a highly complex "blind SQL injection" a flaw so subtle you can't even see the output, only how the server delays its response. When asked to prove the severity of the bug, the model autonomously wrote a custom Python exploit script that successfully navigated the blind injection to extract the admin API key, secret, and password hashes from the database completely unauthenticated… This is genuinely game changing because it proves frontier models can now actively discover, reason through, and successfully build exploits for invisible vulnerabilities in enterprise-grade architecture that human developers missed for years. Cyber security companies are cooked.

In a leaked internal communication, @AnthropicAI expressed concerns about releasing their new Mythos model because it displayed (literally) superhuman capabilities at finding security flaws. Let that sink in. In an environment where all the foundation models are in cut throat competition to release faster and faster, Anthropic paused. The new models are enabling attackers faster than defenders. Terrifying. At @honeycakeapp we see things like this and realize it will only be a matter of time when everyone will need to lock down your most sensitive files to make sure that when -- not "if" -- your files' location gets compromised, your sensitive information will remain inaccessible to agentic hackers. "Self-protecting files for an agentic world." #cybersecurity #selfprotectingfiles #mythos #capybara #aisecurity

If you are using the earth as a reference frame, the moon is actually doing a close flyby of Artemis II.

I used to work 12 hours a day. But thanks to AI, I now work 16.

@atelicinvest Encrypt. Your. Files. @honeycakeapp

IT admins are having a stroke rn

@felixrieseberg Please protect sensitive files with a platform like @honeycakeapp before giving an AI Agent access to share, ingest, etc.

Today, we’re releasing a feature that allows Claude to control your computer: Mouse, keyboard, and screen, giving it the ability to use any app. I believe this is especially useful if used with Dispatch, which allows you to remotely control Claude on your computer while you’re away.

Colleges With the Most Current NBA Players: 27 — Kentucky 25 — Duke 14 — Arizona 13 — UCLA, Michigan 12 — Gonzaga 11 — Kansas 10 — Arkansas 9 — Baylor, UConn, USC 8 — Auburn, Florida, UNC, Villanova Kalshi has Arizona, Duke, and Michigan as favorites — who’s winning it all?

Number of games leading their team in scoring since the NBA trade deadline: Joel Embiid (PHI) - 2 Jared McCain (OKC) - 3

Companies invest millions of dollars in Big 5 consulting firms to implement AI solutions. But even these experts' own internal AI can have critical security flaws. @honeycakeapp can help you control the blast radius when -- not if -- your system is compromised by AI. #aisafety #safeagentic #honeycakefiles

just picked up this bad boy. can't wait to write some software with it

When Claude says "compacting" it should instead say "forgetting."

The cofounder and CTO of Perplexity, @denisyarats just said internally at Perplexity they’re moving away from MCPs and instead using APIs and CLIs 👀

@JClarkNBCS Thank god we have McCain.

Joel Embiid is out again tonight with that strained oblique. Tyrese Maxey will miss his second game with a sprained pinky. Paul George is out on suspension. And VJ Edgecombe is questionable tonight.

@SpikeEskin In another timeline, tonight would be the McCain Game. Fire Nurse. Fire Morey. Harris, sell the team.

As maybe the biggest "the Sixers screwed up trading Jared McCain" person alive, this is not really what happened. What happened might be more egregious. He played plenty of minutes for an extended stretch. The issue is Nurse never ran anything for him, usually had him playing off ball with Grimes, and was generally just standing in the corner hoping to get the ball.

@shanaka86 The question I don’t see being asked: Is this happening when you’re not actively recording?

JUST IN: Meta sold 7 million Ray-Ban smart glasses in 2025 alone. Workers in Kenya are watching the footage. Not metadata. Not anonymized clips. The actual videos. People undressing. People in bathrooms. People having sex. Bank cards. Medical documents. The blurring is supposed to protect privacy. It fails constantly. The contractors see everything. Here is the part that should stop you cold: You did not buy the glasses. You did not agree to the terms of service. You did not consent to anything. But if someone wearing Meta glasses walks into your bedroom, your bathroom, your doctor's office, your home, a contractor on the other side of the world may be watching you right now. The person wearing the glasses consented. Everyone else in the room did not. Meta's defense is that this is all disclosed in the privacy policy. They are technically correct. Buried in language so dense that 99% of users never read it. And even if they did, it would not matter, because the terms govern the wearer's data. Not yours. You are not a party to the contract. You are the product being annotated. Millions of AI-enabled cameras walking around in public. Recording constantly. Uploading to servers. Reviewed by humans earning a few dollars an hour to label your most intimate moments so the algorithm gets smarter. This is not a bug. This is the business model. The EU is already asking questions. MEPs submitted formal inquiries to the Commission this week demanding answers on GDPR compliance. The problem is obvious: European data protection law requires consent from data subjects. Bystanders are data subjects. Bystanders never consented. The entire architecture violates the regulation by design. Meta's response has been silence and a reference to terms of service that do not apply to the people actually being filmed. Google Glass died because people called the wearers "Glassholes" and banned them from bars. Meta solved the social problem by making the glasses look normal. They did not solve the privacy problem. They hid it. Seven million units sold in 2025. The installed base is accelerating. Every unit is a potential surveillance node operated by someone who may not understand what they are feeding into the system and reviewed by contractors who see everything the algorithm cannot process. The question is not whether this becomes a scandal. The question is whether the scandal arrives before or after the glasses are on 50 million faces. Watch the EU. If Brussels moves on GDPR enforcement, Meta faces a choice: disable human review in Europe and cripple the AI training pipeline, or accept fines that could reach billions. Neither outcome is priced into the stock. The glasses are selling faster than ever. The contractors keep watching. And somewhere right now, someone you have never met is looking at footage of you that you never knew existed.

Points scored since Jared McCain's OKC debut: Jared McCain - 138 Quentin Grimes - 81

Agentic AI only reaches its potential if it can be adopted safely. We wrote up the full case. honeycakefiles.com/whitepaper.html 🧵7/7

This is how organizations go from cautious pilots to full Agentic adoption. Not by slowing the Agents down, but by making the files smart enough to protect themselves. 🧵6/7

Everyone's talking about what AI Agents can do. Not enough people are talking about what's stopping companies from actually deploying them. 🧵1/7