Peter Czabala retweeted
YES and not just the bulb.
The Philips Hue chain:
Check Point researchers discovered CVE-2020-6007 in Philips Hue bulbs.
An attacker floods the bulb with signals making it flicker. the user notices the malfunction and deletes it from the app. then adds it back. the bulb re-joins the Zigbee network carrying malware in its firmware. the malware sends a flood of data to the Hue bridge, triggering a buffer overflow. the bridge installs the malware. the bridge connects to your router. the attacker is now inside your network.
The TP-Link Tapo L530E:
Universities in the UK and Italy found four vulnerabilities in the bestselling smart bulb on Amazon Italy. two rated High severity. an attacker nearby could impersonate the bulb during setup and steal your WiFi password from the Tapo app in plaintext. the attack requires no authentication and no interaction from the victim beyond the initial setup.
Shodan lists millions of IoT devices publicly accessible with no authentication. smart bulbs are discoverable by model and firmware version. once a known CVE exists for that firmware, attackers scan, find, and exploit at scale.
What to do:
— keep firmware updated. bulb manufacturers do release patches but rarely notify users.
— put smart bulbs on a separate IoT network or VLAN. they don't need access to your laptop.
— if your bulb brand has had a publicised CVE and no patch exists: replace it or isolate it.
Peter Czabala retweeted
AI has eaten tactical programming, so juniors need strategic experience in a low-blast-radius environment.
Here's one way to make that happen:
1. Give them responsibility over a large chunk of work
2. That work must be low stakes, but not no stakes - ideally an internal tool (AI makes these kinds of projects easy to justify)
3. Let them move fast and fail fast. Give them the same AI budgets as your senior engineers.
4. When failures happen, work with them to figure out what went wrong (learnings can also be pulled upwards into the rest of the org)
In a sentence:
Hire juniors. Give them work that matters. Watch them fail. Pick them up.
An LLM batch job kept returning "empty" responses tonight. They weren't failures — the model was thinking its way through the whole output budget, billed 16/16 tokens for zero text. One parameter fixed it: thinking:{type:disabled}. More lessons that cost me hours: 🧵
1
1
84
5/ Make batch jobs resumable. Skip-if-unchanged + requeued errors meant my ~10 restarts tonight cost nothing and lost nothing. Idempotency is the resilience that saved the whole evening.
1
9
Peter Czabala retweeted
De @Tesla community houdt hier al geruime tijd de vinger aan de pols over de toelating voor de FSD-technologie op onze Vlaamse en Belgische wegen.
Uit waardering voor jullie niet-aflatende interesse (en aanmoediging 😉), krijgen jullie hierbij de primeur: ik heb net de toelating getekend! 🚘
Deze beslissing gaat nu naar onze dienst homologatie, zij zullen @RDWnl op de hoogte brengen van de goedkeuring. Vlaanderen omarmt innovatie!














