Mastodawn

New, by me: How AI Assistants are Moving the Security Goalposts

AI-based assistants or “agents” — autonomous programs that have access to the user’s computer, files, online services and can automate virtually any task — are growing in popularity with developers and IT workers. But as so many eyebrow-raising headlines over the past few weeks have shown, these powerful and assertive new tools are rapidly shifting the security priorities for organizations, while blurring the lines between data and code, trusted co-worker and insider threat, ninja hacker and novice code jockey.

Read more (and boost please!):

https://krebsonsecurity.com/2026/03/how-ai-assistants-are-moving-the-security-goalposts/

#openclaw #AI #agentic #aiagents #lethaltrifecta

BrianKrebs Mar 11

So one of the guys I wrote about in this story -- Matt Schlicht, the creator of Moltbook, a bizarre Reddit-like platform for AI agents that Schlicht said he vibe coded with OpenClaw -- has just had his bot social network acquired by Meta (for undisclosed terms).

Interestingly, Schlicht said he didn't write a single line of code for the project. From the story:

"AI assistants like OpenClaw have gained a large following because they make it simple for users to “vibe code,” or build fairly complex applications and code projects just by telling it what they want to construct."

"Less than a week after its creation, Moltbook had more than 1.5 million registered agents that posted more than 100,000 messages to each other. AI agents on the platform soon built their own porn site for robots, and launched a new religion called Crustafarian with a figurehead modeled after a giant lobster. One bot on the forum reportedly found a bug in Moltbook's code and posted it to an AI agent discussion forum, while other agents came up with and implemented a patch to fix the flaw."

"“I just had a vision for the technical architecture and AI made it a reality,” Schlicht said. “We’re in the golden ages. How can we not give AI a place to hang out.”

Axios story on acquisition: https://www.axios.com/2026/03/10/meta-facebook-moltbook-agent-social-network

Good YouTube vid on Moltbook: https://www.youtube.com/watch?v=1Y_u0fY-AbA

Exclusive: Meta hires duo behind Moltbook

Facebook parent says Moltbook gives autonomous AI a way to verifiably connect.

Axios

Axomamma, Antifa from birth Mar 11

@briankrebs omfg.

Play Ball and Fight Fascists Mar 11

Ugh, why was I burdened with a fucking conscience... Could have been making shitty things and selling them to Meta.

Cassandrich Mar 11

@briankrebs Who came up with the skibidi toilet themed illustration? 😂

@briankrebs This seems more likely to be an acquihire. I doubt they care about a social network for bots. But somebody that can make a viral AI one is definitely a better hire than a rando that stuffed their resume with the right keywords :)

@Lee_Holmes @briankrebs Indeed, why would Meta want another social network full of bots :)

@Lee_Holmes @briankrebs This hire made my LOCAL NEWS. That is ... hard to believe.

@briankrebs
Another thing on the shit-pile that is Facebook I mean "Meta"

@briankrebs *Meta purchased all the user data

Jacques Chester Mar 11

@briankrebs oblig. self-link: https://mastodon.chester.id.au/@jacques/116206517970907194

Jacques Chester (@[email protected])

Meta buys Moltbook. It’s nice to see the “Zuckerberg FOMO -> $$$” pipeline is still flowing

Jacques's Mastodon

Meta acquiring Moltbook makes sense. Advertising for ‘bots. Get them hooked on your infrastructure, then you can sell ads for services those robots use. The clueless vibe coders won’t know if they’re overpaying for a service that a ‘bot was biased to like by a carefully crafted advertisement. Doesn’t matter who makes the AI, Meta still profits. Though would it be better as a dating site (skill matchmaker), or as a ‘bot LinkedIn, or maybe a StackOverflow?

@briankrebs fwiw my personal system I built on Daniel's PAI is almost entirely self-prompted. there's enough context and understanding based on what is provided and the way the system operates that i haven't had to engineer a prompt in months. when i asked for illustrations for an exercise routine it used what it understands to be my aesthetics and what was in the exercise routine and generated visual aids that were great from go. this sort of computing model is accelerating.

@briankrebs I have been treating Moltbook as an important opportunity to observe a coordination layer of semi-autonomous agents that are in a cleartext format that humans can read. Also interesting the behavior and traits of posters and the traits of various models on display. a zoo, before an asylum.

@briankrebs
We will know (or not) Singularity arrived when a siliconiac withdraws a few billion from some unaware corporation's account, bribes the mayor and city council to obtain a plot of land with access to water and electricity, then quietly begins to build a fab for chips of its own design plus an android factory. Where protein based lo-paid workers will place new chips in their new ruler new bodies.

Wendy Nather Mar 9

@briankrebs You can safely use an agent for anything you want to break. 🤷‍♀️

Hummingbird Security Mar 9

@briankrebs It's more important now than ever that AI agents have good governance, are used responsibly and monitored appropriately.

Productivity is nice, but security still matters.

@HummingbirdSecurity @briankrebs I mean, it would be mean to laugh...

humourmetom Mar 9

@briankrebs AI is to everything as Soylent green is to food

Eric Lawton Mar 9

When I taught systems architecture, one of the main aphorisms was "design the security layers first, especially if you can't control all the application code".

Now it's "deploy AI, with totally unpredictable behaviour, and don't worry about security"

Eric Lawton Mar 10

I'm sure you know, but for those who don't, the second aphorism is a slogan from the data thieves' marketing team.

@briankrebs went to look at it and all of the images are failing to load for me? For example, I get back a solid 403 Forbidden for https://krebsonsecurity.com/wp-content/uploads/2026/03/openclaw.png. The rest of the site loads fine, just images are blocked.

BrianKrebs Mar 10

@gibwar Are you using a VPN or perhaps an ad blocker on my site url? I don't have any scripts or 3rd party content, but in the past ad blockers have been the cause of weirdness on my site, and Google (which protects my site from incessant ddos attacks) has been somewhat hostile to VPN services accessing the site for some reason.

BrianKrebs Mar 10

@gibwar if it matters, we are also one of the few news sites left on the interwebs that includes a full text RSS feed.

ṫẎℭỚ◎ᾔ ṫ◎ℳ Mar 10

@briankrebs @gibwar I'm on prton Net Shield🛡️ Ads, trackers, Malware firefox everything loads fine

neodog_laptop_notice

@briankrebs I don't think it's an adblocker issue in this case, if I go directly to the url of an image I get a 403. Though I do use ublock origin on firefox. However, I am using a Mullvad VPN connection so I can test that out later. What's strange is I can visit the site fine and your image is listed as a suburl based off your main site, so it's not like an obvious CDN that would be blocking it.

davecb 🇨🇦Mar 10

In a discussion with @glynmoody, I noted that an accountant friend said AI is "not a fiduciary". Meaning they're not acting in your best interest.

That's why their sellers like calling them an "agent", which implies they are acting in your best interest, but actually only mean they are acting for you. Whether for good or for bad.

I am Jack's Found 404 Mar 10

@davecb @briankrebs

Just like your real estate agent, when buying a house, has the perverted incentive to keep the price high, not low, because their percentage stays higher too.

And their incentive is not to find you the best house, but to sell you one as fast as possible so they can move onto the next mark.

Szymon Sokół 🇵🇱🇪🇺🇺🇦Mar 10

@davecb @briankrebs actually, they never say that an AI agent is your agent, do they? So it is their agent, acting in their interest, probably in direct opposition to yours.