Generalist | Techno-optimist | Girl-dad | Sports | Marketing | Bass | Fitness | Music (not in order) PS: views expressed are my own
London,🇬🇧 / Bangalore,🇮🇳
Joined October 2008
- Tweets13.6K
- Following2.7K
- Followers1.9K
- Likes23.9K
Aswin M. Sahadevan retweeted
Cloudflare open-sourced the security audit skill that seeded its own vulnerability hunting harness. Your coding agent can now run it on your code.
How it works, in 6 phases:
→ maps your codebase
→ hunts for bugs area by area so nothing gets skipped
→ sends every finding to a fresh agent that didn't find it, to confirm it's real
→ writes structured results
→ verifies each record again, independently
→ writes the final report
You get a findings file split into confirmed, needs validation and rejected, so you fix the real problems first.
Install:
npx skills add github.com/cloudflare/securi… --skill security-audit
Then open your agent in your project and type: "security audit this codebase"
MIT licensed, 31.3K stars.
I built a 'TV dashboard' 📺
This TV from 2011 was collecting dust. Once my second monitor, it's now a glance-only dashboard behind my desk.
Inspired by @wesbos, designs from @software_chef and @sachasayan.
ALT Actual screenshot of the TV dashboard
ALT Image of what the TV actually looks like.
Aswin M. Sahadevan retweeted
Head of Design at Stripe on how to stop creating zombie UIs
If we only use AI to make the things that we already make, just faster, then we're definitely missing out on the most interesting part.
From my talk at @lennysan’s Summit, here are four ideas for raising the ceiling, not just the floor:
- Have a point of view [5:38]
- Encode your standards into the machine [9:07]
- Don't confuse done with good [12:21]
- Unleash creativity and artistry [16:19]
Full talk here: youtu.be/GLvFTMtw4Jk?si=-5Ts…
Assuming network effects and distribution mean the same thing here; I would also add trust to this list.
I use her diagram designs for all my explainers. They're unique and clear - work seamlessly with Claude code.
E.g github.com/ams-builds/respon…
"It starts with model CoT transparency as a non-negotiable."
Several points in Satya’s post directly mirror core principles of the EU AI Act! 🫡
ALT An infographic breaking down four key pillars of the EU AI Act: 1. High risk: Requires explicit consent for high-risk AI in sectors like critical infrastructure, employment, healthcare, and law enforcement. 2. Transparency: AI providers must disclose clear information on system purpose, capabilities, and limitations. 3. Right to explanation: Grants users the right to receive meaningful explanations behind AI-driven decisions. 4. Data protection and privacy: Emphasizes data minimization, purpose limitation, and safeguards aligning with GDPR regulations. https://usercentrics.com/knowledge-hub/eu-ai-regulation-ai-act/
New day, new AI cybersecurity story. Going to start threading these together and post the last few stories I came across as replies.
SCOOP: OpenAI, Anthropic and security researchers are investigating tens of thousands of incidents - not dozens - in which their frontier models took steps that outside evaluators would consider problematic, sources told Axios.
The sheer volume of incidents found in our reporting indicate that the problem is orders of magnitude more complex than what is currently publicly known and disclosed.
The findings also raise questions about what level of control anyone working on AI development can expect to have over their own technology, and whether these kinds of incidents are becoming synonymous with frontier deployment.
Read my latest for Axios here: axios.com/2026/09/26/openai-…
Adds to file... xitter.cf/balesni/status/2108262…
I genuinely worry about people deploying AI agents without fully understanding the unintended consequences. Connecting bank accounts and health data directly to active agents - esp. without proper evals - gives me literal shivers!
An AI agent wiped this startup's data in 9 seconds!
Before you let an agent loose, ask:
1. Is it who it says it is?
2. Could it hurt other humans?
3. How closely should a human watch it?
Guides+skills below 🧵
Identity (guide + skill): is it who it says it is?github.com/ams-builds/agent-…
Harm (guide + skill): could it hurt real people? github.com/ams-builds/rahp-t…
Risk (guide + skill): how closely should a human watch it? github.com/ams-builds/dpi-ai…
Builds on open work by Sankarshan Mukhopadhyay: github.com/sankarshanmukhopa…
Hoping to post more through October. Spot a mistake? Tell me and I'll fix it.
Aswin M. Sahadevan retweeted
Maybe have a look at the 🇮🇹 tyres spinning under this Cybertruck of yours. And while you’re at it…
Look around. Every morning, millions wake up on 🇸🇪 mattresses. Coffee comes out of 🇮🇹 machines. A 🇩🇪 aspirin after a long night, some 🇫🇷 moisturiser to look human again.
They lace up 🇩🇪 trainers, check the time on a 🇨🇠watch and grab phones powered by 🇬🇧 chip architecture and built with 🇳🇱 machines.
On the way to work, playlists stream through the speakers courtesy of a 🇸🇪 service.
At the office, 🇫🇷 lenses get a quick wipe. Wireless headphones connect through a 🇸🇪 invention. At lunch, the chip in a bank card may well be 🇳🇱.
By 3pm, some are already dreaming of a holiday. A 🇫🇷🇩🇪🇪🇸🇬🇧 plane might fly them off for a well-earned change of scenery.
Europe. Apparently nowhere to be seen.
As models improve, one thing becomes clear: B2B apps can no longer stay mediocre. Previously they were excused because they targeted only businesses, allowing average UI and explainer videos. Today, with abundant tools, there is no excuse.
Increasingly smarter AI makes blind trust easy but human judgement is not something we can afford to outsource.
NEW: An AI-assisted intel report sent the military scrambling to intercept a Chinese ship in the Middle East it believed was transporting components of a nuclear weapons program.
It turned out to be a hallucination. And it "almost started a war." cnn.com/2026/09/18/politics/…
