Running an army of clankers. 🤖 https://nitter.cf/t.co/aU1P5IqZ4R - AI agents for your site and app dashboard. Qualify, capture, book, support or route.
Cebu City, Central Visayas
Joined October 2018
- Tweets2.1K
- Following134
- Followers267
- Likes5K
Pinned Tweet
I fear it is going to get very bad. We need to rethink how we can defend against this. Local LLMs are getting better so we talk about billions of attacks soon.
Can you please do your job @repasi? Get rid of this notary nonsense and Europe will thrive. You are paid by the people and work for us not the other way around. I hope this is clear.
the German MEP @repasi is quite pro-notary, at the @Europarl_EN @EP_Legal
Even though a lot of data has been shared with him.
e.g. How the 🇩🇪 chamber of notaries threaten notaries and startups who want to digitalize company law.
How they prohibit Europeans from using digital tools (unless they live in Germany)..
How they don’t accept English and cost a fortune etc..
How numerus clausus is outdated and a bottleneck
Time to speak up.
We should not impose feodal red tape and numerus clausus on EU Inc
Christian Gellert 🚀 retweeted
How much reposts and tags for @Alibaba_Qwen @QwenDevs is needed to get me in? 👀
Christian Gellert 🚀 retweeted
Okay now it's Brotli's turn
BREAKING: OpenAI says it found a critical flaw in Brotli, a widely used compression library built into software like Node.js, Bun and Envoy.
The bug could let an attacker take over a system just by sending it malicious compressed data, with no login needed. That works both ways: a hacker could hijack an unpatched server, or a malicious server could hijack an unpatched user’s device.
If you in the market for a fantastic air purifier, buy the AirFanta 5Pro via Amazon and create a review with your experience. Adam is a great guy and the machines are wonderful.
MiniMax M3.1 Flash has been spotted.
MiniMax M3.1 Flash weights are now public. 428B parameters total, 23B activated per token.
Agentic performance: A generational leap in agentic tasks, coding, and cybersecurity — 76.2 on BankerToolBench, 73.7% on DeepSWE v1.1, 61.7 on OSWorld V2, 29.6 on Agents' Last Exam, and 82.2 on CyberGym.
Efficient by design: MiniMax Sparse Attention and DSpark speculative decoding improve inference efficiency.
Hugging Face:
huggingface.co/MiniMaxAI/Min…
ModelScope:
modelscope.cn/models/MiniMax…
API: platform.minimax.io
M Plan: minimax.io/m-planMiniMax
MCode for desktop: agent.minimax.io/download
Oopsy. Not that my LinkedIn bangers circulate freely for everyone now.
wow this is bad!
ENTIRETY of PingCap database used by linkedin, flipkart, airbnb etc. has been copied and stolen due to customer credentials being stored directly in support tickets (LOL)
Attackers ran exfiltration directly on their cloud (LOL) pipeline ran for 22 days, exfiltrating data every 6 hours and NO ONE noticed. Worse, PingCap denies all of it.
Average SRE salary is $300,000 at PingCap btw
What I changed on my own servers after watching these exploit waves:
1. Nothing that does not need to is reachable from the public internet. SSH only over Tailscale. If there is no open port, there is nothing to spray.
2. Keys live in a password manager with fingerprint approval, not in files on disk. I run an agent vault where there is no permission needed for a clanker and the sensitive stuff needs me.
3. Auto updates on everything but with 7 days delay to mitigate supply chain attacks. A patch I install manually will be too late in most of the cases.
It's not perfect. But the mass attacks go for the easy targets first. Don't be the easy target.
Things are going to get very ugly! We need a new approach to defend against nearly any device with CPU being a thread actor soon.
Some stats on #php2shell. In the 10 hours before it was publicly announced, across the @Cloudflare network we saw thousands of attempts to exploit it. That jumped to tens of thousands in the 45 minutes post announcement, hundreds of thousands within 2 hours, and millions now.
I have the feeling that even though I'm in absolute flow right now my context window is running out faster than Opus 5.5's.
How can one offer permanent prompt caching in a way that is reasonable to run as a business? Big move, not sure how they do it.
