Forgot your password?
Close
typodupeerror

Comment Dumped Grok over this (Score -1) 72

Grok was constantly say it was doing something that it had ZERO ability to, and I kept calling it out and it kept apologizing and then immediately doing it again.

As a guy who spend 5 figures a year on Ai, the last thing I want is that. I know Claude and ChatGPT also do it, but Grok was doing it CONSTANTLY.

Submission + - Time To Flash OpenWrt

Submitted by ptorrone
ptorrone writes: The FCC just added all foreign-made consumer routers to its Covered List, banning new models from receiving authorization for import or sale in the US. The agency cited Volt, Flax, and Salt Typhoon cyberattacks, but cybersecurity researchers note the same vulnerabilities existed on American-brand routers too (Salt Typhoon targeted Cisco hardware). No evidence of deliberate backdoors has been presented. Netgear stock jumped 16% after hours. There isn't a single router currently manufactured entirely in the USA. Existing models are grandfathered but firmware updates are only guaranteed through March 2027. If you're running a home network and don't want to wait for the government to pick winners, now's the time to flash OpenWrt, build a router from a Banana Pi or Raspberry Pi with pfSense, or support open-source networking projects like OPNsense. When proprietary supply chains get cut by policy, open-source firmware becomes the supply chain. Adafruit has a writeup with more actionable steps for makers and hackers.

Submission + - The AI CEO's Favorite Book Predicted His Own Crisis (adafruit.com)

Submitted by ptorrone
ptorrone writes: Anthropic CEO Dario Amodei has been recommending Richard Rhodes's "The Making of the Atomic Bomb" to employees for years. Copies sit on coffee tables at Anthropic HQ. The book documents what happened every time Manhattan Project scientists tried to set conditions on how their work was used: Szilard was threatened with internment, Bohr was told he was near "mortal crimes," Oppenheimer learned to stop organizing or lose access. Last week the Pentagon designated Anthropic a supply chain risk for refusing to remove guardrails on autonomous weapons and mass surveillance. OpenAI signed the same contract hours later with the same red lines, just worded differently. Rhodes already told us how this goes. Adafruit has the full review.

Comment REGULATION: the world's worst thing ever (Score -1) 77

by dada21 (#65984450) Attached to: Anthropic Safety Researcher Quits, Warning 'World is in Peril'

Regulators should be afraid of weaponized Ai. So should censors. So should monopolists.

All of the things the State has done in the past 500 years has been corrupt and bureaucratic and caused harm. All. Not most, but all.

All of the people who supported it, from monopolists to lobbyists to activists caused harm.

Ai is undoing it all. Not piece by piece but all at once.

I, for one, can't wait to see folks zapped for restraining voluntary behavior.

Submission + - LLMs Now Write and Hardware-Test Their Own Firmware (adafruit.com)

Submitted by ptorrone
ptorrone writes: Adafruit's Ladyada has been running OpenClaw on a Raspberry Pi 5 with Claude Opus 4.5 managing Codex subagents to autonomously write firmware libraries, then physically verify them against real hardware. The setup parses datasheets, generates a development and test plan, writes the driver, and validates it — using a Neopixel ring to confirm a color sensor actually reads RGB correctly, no humans in the loop. She calls it "agentic test-driven firmware development." The catch: you still need a human reviewing for reward hacking, where models optimize for passing tests rather than correct functionality. Built on top of Adafruit's open-source repos. Blog post and video demo.

Comment hahahahaha -- NO. (Score 1, Interesting) 42

by dada21 (#65958852) Attached to: Unable To Stop AI, SAG-AFTRA Mulls a Studio Tax On Digital Performers

I am a very lame producer, but I make really good money at it. It started as a hobby back in 2015.

I use Ai now with all our customers. We will NEVER hire anyone who has EVER used the word "SAG" in their resume or social media profile.

SAG => monopoly => control who can act => if you don't agree with SAG's politics, you can't act.

It's well past time to remove SAG from the market through market forces. Ai is absolutely AMAZING for my client base, they love being able to change the race, gender, location of a commercial with a few clicks.

I absolutely can't wait to send these SAG clowns packing.

Slashdot Top Deals

MESSAGE ACKNOWLEDGED -- The Pershing II missiles have been launched.

Close