Welcome to the Official Twitter account for #TechTime with Nathan Mumm.
Seattle, WA
Joined May 2020
- Tweets4.7K
- Following20
- Followers690
- Likes10
"SMS of Death": an 8-byte text crashes Android phones, kills calls incl. emergency, public exploit released (BleepingComputer, Oct 31). Android 12/13 never got the fix.
Update today, restart, done.
Eight bytes beats a robot uprising. #TechTimeRadio
Northflank breach: customers told to assume all secrets stolen, AI API keys abused (Wiz, BleepingComputer, Oct 31). Fourth AI-platform breach in Wiz's "Black October."
2.5M stolen Bedrock calls cost one victim ~$100k.
Your API key is cash to somebody. Rotate it. #TechTimeRadio
OpenAI doubled ChatGPT Pro usage limits right after users pushed back on a planned 50% cut. When pricing walks back that fast, the first number was never the real one.
AI hack: screenshot pricing and limit pages before you renew. Memory is cheaper than trust.
#TechTimeRadio
Google, Microsoft, OpenAI and Anthropic committed Oct 30 to one shared AI opt-out: a usage_ai-training line in robots.txt tells all four crawlers no (The Verge). Voluntary, Microsoft starts Feb 2027. But a shared token is how standards begin. Skeptical. #TechTimeRadio
OpenAI asked for $30B, got $11B, cut the round to $20B and dropped the valuation to $1.25T. Now courting family offices (WSJ, Oct 31).
AI hack: when investors discount the leader's own raise, the math changed, not the tech. Watch the money signals, not the demos. #TechTimeRadio
Intel fell nearly 29% Friday, worst day since 1974 (CNBC, Oct 30). Nvidia, AMD, TSMC slid too.
The twist: data center revenue grew 78%. AI wins, PC chips get squeezed.
When AI reorders the supply chain, your next laptop pays the bill.
More: TechTime Radio #TechTimeRadio
Chainalysis says an attacker using "offensive agentic techniques" got customer names and emails (BleepingComputer, Oct 31).
An AI agent hacked the hackers' accountant.
Even security vendors get breached. Verify all alerts.
More: TechTime Radio #TechTimeRadio
A shared ChatGPT link could run attacker code in your session. A math graph's insecure postMessage listener let one malicious link steal private chats and session tokens, patched Oct 26. Shared links are executable content, not text. #TechTimeRadio
Anthropic killed the coding interview. Claude beat most candidates, so they dropped it.
AI hack: if your benchmark can be passed by the tool you're hiring to replace it, test judgment and review skills instead. The interview that tests a machine's skill is over. #TechTimeRadio
Your data can leak without you ever flying compromised. JAL: 92,000 passengers exposed via booking vendor Amadeus, 2,000 with passport numbers. AI hack: inventory what your vendors hold on you; you can't secure data you forgot exists. #TechTimeRadio
An AI coding tool's official extension got trojanized on a plugin marketplace. 5,000 devs installed a persistent SSH backdoor that survives uninstall. Your supply chain now includes your editor's plugin store. Check your authorized_keys. #TechTimeRadio
Reddit is going logged-in only. Search referrals fell 1M+/week as AI answers eat click-throughs, and US daily users dropped ~2.5M in a quarter.
AI hack: when a platform locks its data behind login, it's negotiating with AI companies, not protecting you. #TechTimeRadio
Kimi K3.1 Pro nearly topped the AI leaderboards. Usable today, downloadable Nov 4. Until then it's Moonshot's servers only.
AI hack: promised open weights are a deadline, not a feature. If the date slips, your eval plan slips. Test what you can download today. #TechTimeRadio
$12M in AI chips smuggled in 30 servers, caught at a KL airport. Fifth bust, one syndicate. The chip shortage is so real it built a crime wave.
AI hack: buying discounted GPUs from a reseller? Verify chain of custody. Gray-market chips can arrive locked.
#TechTimeRadio
"Nearly 90 years after War of the Worlds, we're still fighting deepfakes and AI misinformation. Those lessons have never mattered more."
- Nathan Mumm, Episode 320
#TechTimeRadio
Gemini 4 Carbon: Google's new frontier model that finds, exploits AND patches software vulnerabilities. Gated to cyber defenders first, like a weapons export.
AI hack: if your AI can attack code, your test sandbox needs production-grade logging. #TechTimeRadio
OpenAI shipped GPT-6.1 Terra: computer-use AI, 3x faster agents at about half the price. Vendor benchmarks are marketing until independent tests agree.
AI hack: give computer-use agents a test account, never your admin login. #TechTimeRadio
Jobs called touchscreen laptops "ergonomically terrible." Today Apple shipped its first touchscreen MacBook Pro anyway.
Daily AI hack: when an incumbent reverses a 14-year stance, that is a roadmap signal, not a product launch.
#TechTimeRadio
"The definition of buying means ownership."
- Nathan Mumm, Episode 316
#TechTimeRadio
Listening to "320: TechTime Radio: Halloween Gets Real As Cybersecurity Expert Nick Espinosa Joins Us To Explore AI, DNA, Drone Warfare, Election Influence, Cyber Threats, And Human Dependence On Automation | Air Date: 10/27 – 11/2/26" at buzzsprout.com/1136873/episo…
