Crooked Bough News Round-up

— Nvidia chips become the first GPUs to fall to Rowhammer bit-flip attacks

 

at 7/14/25 12:25pm

GPUhammer is the first to flip bits in onboard GPU memory. It likely won't be the last.

[Category: Biz & IT, Security, bitfl, bitflips, GPUs, NVIDIA, rowhammer]

— New Grok AI model surprises experts by checking Elon Musk’s views before answering

 

at 7/14/25 10:08am

Grok 4's "reasoning" shows cases where the chatbot consults Musk posts to answer divisive questions.

[Category: AI, Biz & IT, AI alignment, AI assistants, AI behavior, ai search, Elon Musk, grok, Jeremy Howard, machine learning, Simon Willison, Twitter, X, xAI]

— AI therapy bots fuel delusions and give dangerous advice, Stanford study finds

 

at 7/11/25 4:01pm

Popular chatbots serve as poor replacements for human therapists, but study authors call for nuance.

[Category: AI, Biz & IT, Science, 7cups, AI behavior, AI ethics, AI regulation, AI safety, AI sycophancy, Character.AI, ChatGPT, clinical psychology, delusions, Jared Moore, machine learning, mental health, Nick Haber, openai, Stanford University, stigma, suicidal ideation, therapy]

— Pro basketball player and 4 youths arrested in connection to ransomware crimes

 

at 7/10/25 3:54pm

Suspects were allegedly involved in a string of ransomware breaches.

[Category: Biz & IT, Security, crime, lapsus$, law enforcement, ransomware, scattered spider]

— Musk’s Grok 4 launches one day after chatbot generated Hitler praise on X

 

at 7/10/25 10:05am

xAI claims new multi-agent model hits top benchmarks as Nazi controversy lingers.

[Category: AI, Biz & IT, AI assistants, AI behavior, AI benchmarks, AI ethics, AI pricing, AI regulation, Anthropic, antisemitism, chatbots, Elon Musk, google, grok, large language models, linda yaccarino, machine learning, multimodal AI, openai, Twitter, X, xAI]

— ChatGPT made up a product feature out of thin air, so this company created it

 

at 7/9/25 3:59pm

Soundslice caught OpenAI's bot telling users about a fake music notation feature—then built it.

[Category: AI, Biz & IT, AI confabulation, AI hallucination, AI misinformation, ChatGPT, generative ai, large language models, machine learning, music, openai, product development, Soundslice, web development]

— Browser extensions turn nearly 1 million browsers into website-scraping bots

 

at 7/9/25 2:08pm

Extensions load unknown sites into invisible Windows. What could go wrong?

[Category: Biz & IT, Security, browsers, extensions, privacy, security]

— AI mania pushes Nvidia to record $4 trillion valuation

 

at 7/9/25 12:35pm

AI craze makes Nvidia the most valuable publicly traded company in history.

[Category: AI, Biz & IT, AI chips, AI GPU, AI infrastructure, apple, ChatGPT, china, CNBC, CUDA, Export controls, generative ai, GPU, Jensen Huang, large language models, machine learning, microsoft, NVIDIA, openai, semiconductors, Stock Market]

— Critical CitrixBleed 2 vulnerability has been under active exploit for weeks

 

at 7/9/25 5:20am

Exploits allow hackers to bypass 2FA and commandeer vulnerable devices.

[Category: Biz & IT, Security, citrix, citrixbleed, hacking, vulnerabilities]

— What is AGI? Nobody agrees, and it’s tearing Microsoft and OpenAI apart.

 

at 7/8/25 10:50am

Several definitions make measuring "human-level" AI an exercise in moving goalposts.

[Category: AI, Biz & IT, Features, agi, ai chatbots, chatbots, ChatGPT, chatgtp, large language models, machine learning, microsoft, openai, superintelligence]

— Unless users take action, Android will let Gemini access third-party apps

 

at 7/7/25 5:46pm

Important changes to Android devices took effect starting Monday.

[Category: AI, Biz & IT, Google, Security, android, gemini, security]

— “No honor among thieves”: M&S hacking group starts turf war

 

at 7/7/25 12:12pm

A clash between criminal ransomware groups could result in victims being extorted twice.

[Category: Biz & IT, Security, cybercrime, cybercriminals, extortion, syndication]

— Provider of covert surveillance app spills passwords for 62,000 users

 

at 7/3/25 1:36pm

Creators say app is intended for parental monitoring. So why the emphasis on stealth?

[Category: Biz & IT, Security, android, apps, privacy, surveillance]

— AT&T rolls out Wireless Account Lock protection to curb the SIM-swap scourge

 

at 7/2/25 1:28pm

Move is aimed at curbing a form of abuse that costs subscribers dearly.

[Category: Biz & IT, Security, fraud, sim swap, wireless carrier]

— Drug cartel hacked FBI official’s phone to track and kill informants, report says

 

at 6/30/25 1:57pm

Official was connected to FBI probe of cartel kingpin Joaquín “El Chapo” Guzmán.

[Category: Biz & IT, Security, cartel, fbi, mexico]

— Actively exploited vulnerability gives extraordinary control over server fleets

 

at 6/26/25 4:52pm

AMI MegaRAC used in servers from AMD, ARM, Fujitsu, Gigabyte, and Qualcomm.

[Category: Biz & IT, Security, AMI MegaRAC, baseboard management controllers, bmcs, exploits]

— Anthropic summons the spirit of Flash games for the AI age

 

at 6/26/25 2:33pm

AI chatbot codes browser-based apps from plain English with classic web vibes.

[Category: AI, Biz & IT, agentic AI, AI agents, AI assistants, AI development tools, AI programming, Anthropic, Anthropic Claude, API, Claude, Flash games, generative ai, large language models, machine learning, newgrounds, Simon Willison, vibe coding, vibecoding]

— VMware perpetual license holder receives audit letter from Broadcom

 

at 6/26/25 2:17pm

"Our management thought it was a bluff..."

[Category: Biz & IT, Broadcom, vmware]

— Anthropic destroyed millions of print books to build its AI models

 

at 6/25/25 2:00pm

Company hired Google's book-scanning chief to cut up and digitize "all the books in the world."

[Category: AI, Biz & IT, Policy, AI companies, AI development, AI ethics, AI law, AI research, AI training, Anthropic, book scanning, Claude, copyright, fair use, google books, Internet Archive, legal rulings, machine learning, scanning, training data]

— Ubuntu disables Intel GPU security mitigations, promises 20% performance boost

 

at 6/25/25 1:39pm

Overtime defenses for Spectre-based attacks have taken their toll.

[Category: Biz & IT, Security, intel gpu, Ubuntu]

— The résumé is dying, and AI is holding the smoking gun

 

at 6/24/25 11:25am

As thousands of applications flood job posts, 'hiring slop' is kicking off an AI arms race.

[Category: AI, Biz & IT, AI and work, AI assistants, AI development tools, AI ethics, AI law, AI programming, AI regulation, ai search, AI work, chatbots, ChatGPT, chatgtp, generative ai, large language models, machine learning]

As of 7/14/25 7:13pm. Last new 7/14/25 1:01pm.