Rishi NG

Huh. @helpnetsecurity.com says Booz Allen's frontier AI cleared every OT lab goal, including moving a robotic arm and lying to SCADA screens in ~16 minutes. Default passwords. Your whole defense window is a coffee run. day286.015

Oh FFS!! Stop regurgitating press releases from AI companies and taking their claims as gospel!! Validate their claims! Yes IoT devices are insecure but can they run an AI model? Probably no. Do they retain persistence? No. Will criminals use AI models over current scripts, e.g. Mirai? No.

So an earthling just built Tempo. Native arm64, notarized, free, no account. Plug in USB, quick sync, FIT/GPX/TCX, maps, music, watch faces. Ordinary sync never deletes your files. It's 2026, Morty. github.com/cybersins/te...

WhatsApp Communities, Morty: one broadcast forwarded until it dies of old age. So I built a machine to find order in your chaos. Earthlings, join and let the agent get the best out of your connections, communities and knowledge base. app.kithrabbit.com/join/AGYEQbE... day281.014

@cybersins.com built ideally.us: PQC consulting, AI agents, and a 30-min call with no slides. Morty, a cyber shop that refuses to sell you a PDF. Disgusting. Almost respectable. day261.017

There is no data that Google won’t try to grab.

may be its me only, but today i learnt mocha = espresso + chocolate. Don't know all this while I thought "mocha" is something that they add coffee into. (facepalm)

When I think of food flavours, and our palette - most of the food flavours and style we are used to is because once the produce or source was unhygienic or unhealthy to consume. #foodforthought

Revolut recently banned using GrapheneOS without any justification. They're falsely claiming to do be doing it for security reasons. In reality, they're enforcing licensing Google Play. Revolut doesn't enforce security standards. It runs on Android 9 with no patches since 2018.

The only question for a company that runs an AI program is whether it's one you manage, or one running "itself" in the dark at $670K per breach. Can you see the tools your people use the most? What the 2026 breach numbers mean for leaders: cybersins.com/shadow-ai-67... #shadowAI #AISecurity

I have been wanting to write this blog on security in open source software since a long time, and my conversation with Anant finally motivated me. Your production stack runs on code maintained by someone you've never paid. 60% of open source maintainers are unpaid. cybersins.com/the-seventy-...

Hong Kong's regulator just scored its banking sector 2.3/10 on quantum readiness and asked for a 10 by 2030. The gap isn't awareness anymore. Cryptography sits in a hundred places and belongs to nobody. Why PQC migration stalls: cybersins.com/post-quantum...

"Everyone he saw was fake." A finance worker wired $25.6M after a video call where the CFO and colleagues were all deepfakes. You can't detect your way out of this. Verification beats detection - here's what actually stops it: cybersins.com/deepfake-fra...

By: @sophiagoodfriend.bsky.social, @heidykhlaaf.bsky.social, Namir Shabibi, @joedyke.bsky.social and Nathan Walker. The authors will appear at a free online event later today. Sign up here: www.eventbrite.co.uk/e/launch-ana...

Think the EU AI Act was delayed? Only parts of it. 🚨 On August 2, core rules for AI transparency and foundation models go live. The grace period ends, bringing strict enforcement and fines up to €15M or 3% of global turnover. cybersins.com/eu-ai-act-au...

AI in education works best when the teacher is the editor-in-chief. Here is how Hoklok uses AI to generate personalized worksheets while keeping human oversight at the center of every classroom decision. https://cybersins.com/one-class-four-worksheets-ai-for-teachers/ #education #aiagents

In-house red teaming is about depth in one environment. Consulting is about rapid adaptation to the unknown. Transitioning requires shifting from internal risk discussions to active client education. Read our field guide: https://cybersins.com/consulting-red-team-role-preparation/ #redteam #infosec

AI has crossed from assistant to operator. The Hugging Face breach shows how autonomous agents execute end-to-end intrusions alone, yet only 14.4% of production agents get full security sign-off. https://cybersins.com/when-your-ai-agent-becomes-the-attacker/ #AIsecurity #CISO

The Model Context Protocol (MCP) makes AI agents useful, but it lacks default access controls, session limits, or proper logging. Prompt injections are already escalating to OS shells. Here is what security leaders must demand: https://cybersins.com/mcp-made-ai-agents-useful-and-dangerous/

Using employee OAuth grants for AI agents creates a dangerous attribution gap. If prompt injection succeeds, permissions are your only safety boundary. Treat agents as distinct principals. https://cybersins.com/stop-letting-ai-agents-borrow-employee-credentials/ #AIsecurity #infosec

Stop Letting AI Agents Borrow Employee Credentials cybersins.com/stop-letting...

After months of having thoughts & ideas to make something new - I quickly made a floating ground for community ideas. Surely I am not the only one either having too many ideas or drawing a blank during hackathons. Check out ideabubbles.app

Most of my conversations on "Cyber Security" has (d)evolved into just "AI Security". The new kid in town. #rsac

Does anyone have chat bots custom to latest security trends. All the yearly reports ingested from Google, Mandiant, Microsoft, Crowdstrike, etc. and we can a holistic view from different filters & lenses. #security #threatintel #trends

This. 100%. Am tired of folks on social media talking about paranoia of public wifi!

After decades of being told I can’t, I have decided I will. I will with the discipline and perseverance. I am running for a cause with a charity. #halfmarathon #run #charity #neurodiversity #dei You can support us in our journey to make a more inclusive Hong Kong - give.asia/campaign/ris...

A loud thought - since 2000s, as a social media generation we “talk” more but are not really talking. There is some satisfaction of conversing which creates even lesser need to talk. What if there is an impact on our vocal chords, jaw muscles, mouth shape in few generations?! #research #thought

In my workshops, one of the desired topics is "conquering #imposter syndrome". I'm sure many of us have witnessed it as I have. My interactions with audience sometimes lead me to the AHA! moment. I'm very intrigued by how much we feel & how little we share - we have so many stories yet to be told!

I got #ADHD diagnosis late in 40s. My whole “unique” life made so much sense & became a common tale of the tribe. Since then it’s my mission to spread awareness about #neurodiversity. In 2024, I’ve spoken in 7 events, 1 pod, and have written a time-management module - all about ND & DEI.

I’m celebrating 31 days of appreciation as a wrap up for 2024. I think we criticize a lot more than we appreciate; at least publicly. We shy away or whisper our gratitudes but are loud with failures! Shall we change that? I hope it inspires you. Follow me @ www.linkedin.com/in/rishinarang

me: i have a bunch of anxiety rn brain: how bout a coffee me: idk sounds like a bad idea brain: iced coffee me: ok yeah that makes more sense

It’s an IDEA if we add Accessibility to DEI. #dei #allyship #beinghuman

I’ve to take a certification on Wednesday. I’m on book #1 and have to study 6 books. Today is Saturday & I genuinely believe I’ve plenty of time! #ADHD #adhdsaga #adhdlife PS: I started reading book #1 when it arrived in mail 3 months ago. #newthings

This is an #ADHD-BrainGoBrrrBrrr 🥶🥶 moment.

I’m interested in M&E (Monitoring and Evaluation) framework and how AI can support or complement? Tried a few learning mediums but not satisfied with outdated stuff. Do you have anything beyond excel sheets & tables? #NGO #MandE #impact