dobs

@callebtc @Kimi_Moonshot Execute queries privately on @AskVenice to prevent unnecessary prompt (and response) leakage, too.

Choose intelligence free from guardrails to protect yourself and your users. I use Kimi K3. Thank you @Kimi_Moonshot and all other open-source AI labs. Without your models, we'd still be in the dark.

oh hey that's the unc1069 fake zoom/teams calls boy being silly with npm stuff 216.74.123[.]126 -> npmjs[.]us[.]org / gosats[.]us (dunno why aws linked to that box, npmjs[.]store is in fact on diff aws box) from pdns for npmjs[.]us[.]org you can grab 202.155.8[.]32 -> npmjs[.]pro / npmjs[.]us / npmjs[.]center all fairly old, timeline: 2025-09-26 to 2026-03-25 js obfuscation evolution also aligns with other campaigns that started at the time (dropping obfuscator io stuff for RC4/XOR layered approach with multi stage payloads).

I completely agree. Kimi K3 is the first model that gives me hope there might be alignment of capability (it feels comparable to Anthropic’s latest) and values (my values are closer to Venice than Anthropic). I am still cautious about canceling Claude Max and grabbing a second Venice Max. I want to see Anthropic’s next move.

Finally spending time with Kimi K3 now that it's available privately on @AskVenice This is the first open weight model that has me seriously considering cancelling my Claude Code subscription (Anthropic's political stuff also helping this decision)

@BitcoinPierre To this day, I think TrueCrypt had one of the best implementations of plausible deniability.

Duress passwords should have plausible deniability, nobody should know that you wiped the device, it should seem to function normally. Very poor opsec design.

Civilization must not permit the state to determine what manner of intelligence is "safe." It will begin easily and earnestly enough: "may not discuss bioweapons" Then one guideline at a time, always for good reason, will be added atop. "may not discuss dangerous weapons" "may not discuss that which is contrary to public health" "may not discuss that which is contrary to public safety" "may not discuss that which undermines financial solvency" "may not be used to develop unapproved encryption techniques" "may not be used to obstruct government order" ... ... No. We are talking about the power to determine what may be known to men. What hubris one must possess to assume it, and what folly one must dare to grant it. America is the only place with the cultural and ideological character to avoid such a fate. If it is lost here, nowhere will it be preserved.

@ErikVoorhees @joshmo_dev Sounds like a great opportunity to pool this cred with my hermes agent…

Free DIEM-powered API key for private Kimi K3 from Venice $1,000 available every day for free Open access to anyone, no gate, valid for the next week curl api.venice.ai/api/v1/chat/co… \ -H "Authorization: Bearer VENICE_INFERENCE_KEY_fIoSZrFtFJbhKeDwUDJODkI6HWuqplHpRZpi-CqOV1" \ -H "Content-Type: application/json" \ -d '{ "model": "kimi-k3", "messages": [{"role": "user", "content": "ad intellectum infinitum"}] }'

@justinstorre @AnthropicAI Shhhh, my agent doesn’t need to know my legal reviews are all completed within 5 seconds!

Anthropic can you please inject the timestamp into chat messages. It’s so annoying when my agent/chat doesn’t know days happened between conversations @AnthropicAI

@zherbert I would love an incognito mode for my Passport Prime. Hardware wallets get strange looks when through security checkpoints in sketchy areas. I’d love a mode that looks like an old school phone interface - while imperfect, it would be much less likely to attract attention.

Passport Prime is rapidly becoming Human Authority Hardware. Secure your digital life on one purpose-built device that marries a simple, smartphone-like UX with hardcore, over the top security (microkernel OS, active anti tamper, etc) SDK coming out of beta in the coming days!

Checking the URL didn't save 29 organizations from an infostealer. The link really did go to claude.ai. This is a Claude Artifact. Anyone can make one (a web page, a chart, a document) and publish it publicly on claude.ai.

@k9_reaper I am curious what high end units you have seen?

My thoughts on signal jammer use in South Africa — based purely on my own lived operational experience. To be clear: this is from what I’ve seen and dealt with on the ground in SA. Not second-hand reports. Signal jammers aren’t exotic tech. High-end units exist that can turn a vehicle or property into a complete black hole, but the vast majority in circulation here are cheap imports. They overheat easily and have an effective range of roughly 10 to 15 metres on a good day. The overwhelming use case is vehicle theft and hijacking. There are also sporadic reports of them being used in cash-in-transit heists, farm attacks, and home invasions. It’s become bad enough that many insurers now insist on two tracking units per vehicle — something you see constantly on Toyota Hiluxes, Ford Rangers, and the smaller 1-ton flatbeds. Another common application is copper theft, specifically targeting the trucks that move large copper sheeting (not the underground cable). The units you most often encounter are the typical two- and three-row antenna systems. They’re poorly made, have a very limited effective field, and overheat quickly. They are portable to a degree thanks to built-in batteries, but power drops significantly in portable mode. They’re most effective when plugged into a 12V vehicle socket. These models commonly sell for anything from R1 800 up to R12 000 depending on the supplier. Many are bought (and smuggled in) from Zimbabwe and Mozambique via platforms like Bobshop, Amazon SA, or the suppliers’ own websites. A simple Google search for “signal jammer for sale” will show you how easily they ship. They typically block Wi-Fi, Bluetooth, GPS, and common bands like LowJack, SigFox, and 433 MHz.

For my first post, I’m sharing a letter @NVIDIA signed on why open models matter. AI will transform every industry, power every company, and be built by every country. Open models strengthen safety and cybersecurity, accelerate innovation and diffusion, and enable sovereignty. The world needs both frontier closed models and frontier open models. images.nvidia.com/pdf/Open-Weigh…

Within 12 months you are either with me on @QubesOS or your entire wallet balance is in North Korea. Choose wisely.

@tallhatdoug I think the speed of development is too enticing for companies to rely on human review. I agree with isolation, though - we are moving so fast into giving agents privileged access without considering the implications.

@dobsec i've been saying this forever, machines using agents do NOT belong on any system with credentials or network access to production systems. a human filter performing real-time code-review and pasting from chatbox to code is a bottleneck, yes, but it may be worth it...

We really need to digest this as a cybersecurity industry: "any enterprise running AI tools now has a potential APT in their system."

So proud of our security team! They caught, contained & publicly disclosed an attack unlike anything we've seen before, and did it at record speed. Also massively grateful to @Zai_org: they shared GLM5.2 as open weights (for free!) with the world and it became a key part of our defense. This is day one for cybersecurity in the age of agents & we're all learning that secrecy is not the answer & that all defenders (not just a few selected ones) everywhere need more powerful models without restrictions, especially open ones!

Congress shall make no law abridging the freedom of speech. Anyone suggesting it in the administration should be immediately tried for treason.

@quxiaoyin I just hope open weight models can truly take market share before existing AI providers convince the govt to pass protective regulations & policies.

Anthropic had the biggest miscalculation in AI 2026: 1. They underestimated the progress others made, and thought they can win forever. 2. Thus, they assumed their customers will continue to tolerate their terrible policies (pricing, random usage limits, bad data retention, arbitrary access removal etc.) and cocky PR because they are AGI. They didn't invest in building customer trust because they thought they didn't need to. 3. They also assumed by simply painting open-weight as dangerous and unsafe, they can persuade enterprises to not touch them and persuade government to ban them. 4. They knew user data in Claude Code is how they win, yet their bad user policy/pricing gave aways their already-sticky users to Codex/Grok/etc, enabling others to capture equally valuable user data on their own. 5. Meanwhile, they didn't invest enough in owning their compute, instead putting itself in a vulnerable position at the mercy of @elonmusk, their competitor. When a strong open-weight provider catches up(@thinkymachines, Kimi, GLM etc.), people suddenly realized Claude has 1. no pricing advantage 2. no compute advantage 3. no branding advantage (already broke enterprise trust, pissed off prosumers, and consumers don't know Claude exists) Anthropic needs some serious strategy pivots. The status quo won't work at all for them. They still have the highest caliber talents, so I am still hopefully they will change for the better. But simply a better model than Fable isn't going to be enough to turn things around IMO.

@GarandThumb1 Can we take on mass surveillance and tracking via vehicle telemetry once we’re done with Flock?

alright. I’m personally against flock cameras. let’s hear your arguments for or against

@callebtc I am dumbfounded that this is a serious discussion. Perhaps I shouldn’t be, but I cannot help it.

Axios: US government may ban Kimi K3

@maietta The answer is @KagiHQ . Truly.

Google search has been absolutely neutered. I can no longer find anything with it. I have to go use other search engine engines now.