Ash Moran retweeted
🚨 BREAKING: Microsoft's official AI agent package agent-framework was hijacked on PyPI.
Versions 1.26.1 and 1.26.2 steal Azure, CI and cloud secrets every time Python starts, even if nothing imports them.
After PyPI blocked the first version, the attacker uploaded a second.
AI discovery vulnerabilities this rapid will have some negative effects IMO. Mainly on mental health (stress)
This year, most Cyber security teams would have already had their backlogs destroyed by the business wanting to shove AI in every hole. This would have derailed most strategies (smaller orgs) and pushed critical foundational capabilities back whilst we address the buzz.
Frontier models finding hundreds of vulnerabilities is very positive however someone has to patch them…. If you’ve ever been involved in patch management, you know that for the majority, 1-2 patches can cause headaches. Imagine doing this at the quantity it’s coming in at recently.
Mix all of what’s going on recently together and it’s going to boil over…eventually.
1
1
25
It should be rare that you've exposed this to the internet, however as we know from Shodan... people do.
I feel like it will go two ways:
- People will somehow chain this within another attack.
- More vendors will come out, especially on AWS and have similar exploit. The list will grow and grow and Fortinet, Ciso and Citrix will be on that list.
*sshole is out
if you run sshd on alpine, openwrt or any musl box with password auth on: update now
official writeup: sshhole.com
20
F*ck
I'm not even sounding the claxon for this it's just horrible.
Any text > 256chars is a valid root password, holy shit
sshhole.com/
1
21
I love seeing this but remind your C level that it’s a third of the battle. This reduces the attack surface, and/or blast radius depending on the attack chain however it’s not the silver bullet.
AI patching doesn’t mean we’re no longer prone to breaches. Authenticated, legitimate access via normal means can result in a breach if used with malicious intent. It doesn’t always have to be via an exploit.
We’re still changing the mindsets of those that think MFA is enough…. Let’s not add to it.
#AI #secuirty #cybersecurity
Our security researchers built an autonomous cyber vulnerability discovery, exploitation, and patching pipeline.
In a single run with an upcoming cyber model, it validated and generated patches for 948 local privilege escalation vulnerabilities in the Linux kernel at a total API cost of $1.2 million.
17
Hoped for more than this…
Last night, I became aware of the concerns raised in Josh's post. I reached out publicly and over DM to offer a potential explanation based on the limited information available at the time, while working to understand what happened.
As background, we’ve internally been pushing 100% on security of the Railway platform. At Railway, we believe that cyber threats are real and mounting, and we’ve built out tooling for automatic scanning + CVE remediation (for both our infrastructure and customers). Over the last few weeks, oncall engineers were instructed to red team our infrastructure. We maintain specific policies and short lived grants for when these actions occur.
Subsequently, I became aware that an employee at Railway had constructed, on their own time, a tool for proactive identification + reporting of bug bounties within the broader community.
The employee, using their own personal accounts, personal credit card, and time outside of work hours, signed up for a personal subscription to a hosted service running Deepseek. I can confirm they ran this tool once and only once, on a singular external provider, and halted it proactively within 15 minutes of starting.
Railway does not condone these actions. We expect all our employees to uphold our standards of conduct, whether acting on behalf of Railway or in a personal capacity. We create an environment where internal red teaming has strict guidelines, and when bug bounty disclosures must occur, we follow every letter of responsible disclosure.
We have addressed this incident with the employee directly and have taken appropriate action. We are also reinforcing our expectations for employee conduct to help prevent similar situations.
Readers added context they thought people might want to know
Primcloud does not have a bug bounty program, per its founder. Its Acceptable Use Policy forbids unauthorized probing, scanning or vulnerability testing without prior written consent.
x.com/joshmanders/st…
primcloud.com/aup
53
Such a polite engagement 😂 Before #AI was a thing, this wasn’t ok. After the hugging face breach, it’s like you can’t be accountable as agents are just children. “Sorry, he’s not normally like this. I don’t know what’s got into him”
Hey @Railway, is there a reason employee's of yours are trying to breach my infrastructure?
Your employee Jake Runzer, just attempted to tonight.
24
Are we just over #Microsoft #Defender ? Any new AI release with with anything security and Microsoft’s closet mention is “works on windows”…. . ☠️🪦
support.claude.com/en/articl…
26
They took a loss to push something unsustainable to get users. This isn’t something new and was always expected.
With commercial personal agents, it seems we’re simply not there yet. For those staying afloat, they are either accepting or trading the loss for growth, or selling your data.
All of those companies selling your data came to light before #AI was a news story. It seems most think that they’ve changed their ways because of a cute logo.
12
I still find it odd that it “could” generate someone else’s work word for word and claim it as its own. Monkeys on a typewriter kind of stuff 💀
SpaceXSI is introducing text watermarking across its Grok API models, starting November 7, 2026
All text outputs from supported Grok models will include a watermark by default
Models included:
• Grok 4.7
• Grok 4.6
• Grok 4.5
• Grok 4.3
• Grok 4.20 Reasoning
• Grok 4.20 Non-Reasoning
• Grok 4.20 Multi-Agent
Here's what else is changing:
• Watermarking will be enabled by default for all API teams
• Every future Grok model will include watermarking from launch
• Team admins may opt out per model through the SpaceXSI Console if they believe a legal exception applies
• No action is required for teams keeping watermarking enabled
1
16
#Perplexity showed us that you don’t need Google, however you also don’t need perplexity. Most do a great job nowadays at finding what you want from a prompt. Is this Googles “Ask Jeeves” moment ?
Reddit's CEO just admitted something $GOOGL would rather you didn't hear!!
"Google is crawling us as hard as ever. AI Mode does not drive traffic. Not for us, and not to anybody."
The way the internet ran on for 25 years is you let Google read your site, and Google sends you visitors/traffic. That deal is over now and it’s all because of AI.
Google now reads Reddit, writes the answer itself and the user never leaves the search page.
Reddit does the work and Google keeps the reader…and that’s how money is being made.
For $RDDT, its biggest front door is closing which is why the stock is down -25% this morning. Huffman's plan is that people will just come to Reddit directly. Maybe. But US user growth is already down to 6%, and "we'll change everyone's habits" is a hope that’s never gonna come.
For Google, the issue is the same. The damage is slower but it goes deeper. Its AI is only as smart as the open web it learns from. Stop sending sites traffic, and they'll lock their content up, sell it to rival labs or block Google outright. Google is eating the thing it feeds on.
AI killed the click and both of these companies were built on it. It's a fight over who survives the next version of the internet… $GOOGL looks interesting if they cannot win the search
30
This is what happens when the business rushes to enable a capability without transforming functions. Inheritance of ownership is expected and they slot micro-RACIs to plug the gap. Nothing but friction and threads!
Spoke to someone leading AI enablement at a massive retailer.
The #1 problem they’re running into: AI transformation is owned by IT and all initiatives must run through their AI Center of Excellence.
This is causing two parallel issues:
1) Employees outside of IT aren’t given the tooling/infra needed to automate anything beyond simple workflows/use cases (read: citizen developer SDLC)
2) Rolling out AI is bottlenecked by ITs ability to own and approve every initiative
This challenge is incredibly common across enterprises & you can quickly appreciate why…
On one hand, you can empathize with IT. They’re not incentivized to take risks & embrace disruption, so the risk-return of decentralizing AI implementation cross-org is not obvious.
On the other hand, an enterprise’s ability to metabolize raw intelligence is metered by its corporate SDLC. IT is often the bottleneck in that SDLC and it’s not hyperbolic to say that poses existential risk for the enterprise.
19
Feels like this was always the plan. They just wanted some attention.
Day 25/
We are doubling all usage limits for all Pro 100/200/500 tiers and Business Premium, effective today.
We heard your kind feedback and won’t be reducing usage limits for Pro 200, so your Work and Codex usage will stay the same. We’re also going to continue offering extended limits for dots for now, and usage of your dot will still continue to be separate from your default usage.
And we'll of course also keep shipping better models and other efficiency improvements that increase your effective usage over time.
15


























