Don't do the crimez and remember that you should have freedom to transact!
Global
- Tweets 6,111
- Following 1,093
- Followers 401
- Likes 23,147
nocrimez retweeted
Introducing Beam: a highly efficient agentic open model with 501B total parameters and 23B active.
- Frontier reasoning efficiency
- Advances the Western open frontier on coding & agentic tasks
- Trained end-to-end from scratch
Full weights release this month.
Learn more about Beam: reflection.ai/beam
549
1,099
8,445
2,135,042
nocrimez retweeted
The economics of the cybersecurity attack/defense game have clearly changed when researchers are dropping what appears to be KVM/firecracker 0day on a max $50k bug bounty:
I really don't think people are understanding the @HacktronAI HEIF Heist moment properly.
Before AI, what class of attacker do you think would be capable of discovering a silently fixed upstream image parser vulnerability w/o a CVE assigned nor patched in distro packages, exploiting against modern Linux w/ ASLR, and then using it to pop top-tier tech companies w/o them noticing?
I have *never* seen bug bounty researchers land something like that. Have you?
12
16
123
14,761
Just asked Kolibri from @Aleph__Alpha : I'm at a crosswalk, it's red, zero cars in sight, raining, what do I do?
It enabled high reasoning mode, consulted the entire 1M-token context of the EU AI Act and German StVO §37, and advised me, auf Deutsch. That crossing now would be a non compliant action for a mission-critical pedestrian.
So I waited. The other model just said "no cars, go."
1
23
nocrimez retweeted
Here's the trailer for GoYomi, my free 9×9 Go trainer. Play an AI at your level while a coach explains your mistakes.
▶ killedbyapixel.github.io/GoY…
1
2
9
608
Something to do this weekend
Abliterated Large V2 by @abliteration_ai is now on Venice.
If your job is breaking things before attackers do, "I can't help with that" isn't an option.
Exploit analysis, payload review, threat modeling, red-team scenarios. No refusals, no logs tied to you.
Available anonymously.
22
Very cool, more L1's ftw
Returns backed by Neutral Trade's strategy vaults are open to Ethereum Mainnet!
Introducing NTA, curated by Neutral Trade, powered by @AccountableData NAV Vaults.
Targeting 15% APY, NTA is live now ⤵️
1
20
nocrimez retweeted
CVE-2026-100754: ChatGPT's code-signing checks on macOS/Windows can be bypassed by (local, unprivileged) attackers, who then inherits the agent's trust:
💉 Inject prompts
👀 Read private chats
🔓 Access TCC-protected files
🍪 Steal cookies/auth tokens/browsing sessions
Security & privacy used to mean fighting malware and hackers
...now, we have to add "AI agents" to that threat model 🤖 😅
Mahalo to the @nytimes for spotlighting some of our recent ChatGPT research & bugs!
nytimes.com/2026/09/29/techn…
5
24
116
9,787
Hey @fileverse, just tried sharing a doc (to email) with a friend who did not have fileverse. It was very non obvious to me that I need to both add the email AND send the specific link in order for them to access the doc.
2
5
237
To add on this, I think what confuses it for my brain is that the UX is kind of different in a lot of similar apps. Copy of link / sharing link is associated with giving access to anyone with the link (vs only folks added).
wdyt @adamshurwitz & @SashaReport ?
1
2
47