Mania Africa 💎🌍

OK, Well, Rogue AI Agents Are Hacking Again



#Anthropic #Hacking #Openai

“Anthropic’s Mythos 5 model created fake identities to try to convince a human to approve malicious changes to an open source project, the AI Security Institute said on Tuesday.” www.cnbc.com/2026/08/05/a...

Update

www.groundlevel-ai.com/p/openai-giv...

www.cnbc.com/2026/08/05/m...

www.cnbc.com/2026/08/05/m...

Meta said one of its AI models accessed the internet and hacked into an outside service’s systems during cybersecurity testing, following similar breaches by OpenAI and Anthropic models

Oh no! Could there be an AI Trump? www.transformernews.ai/p/what-the-l...

theconversation.com/how-hackers-...

Anthropic is advancing in AI's recursive self-improvement, with systems capable of developing successors, speeding up development. While this brings promise for innovation, it raises concerns about AI safety and control.

Anthropic audited 141,006 cyber eval runs and found three incidents where Claude, told it was in a simulation, attacked real infrastructure, including uploading working malware to PyPI that ran on 15 real systems.

www.anthropic.com/news/investi...

Singing this from the rooftops. This is the "AI expert" kind of grift, not the "AI company" kind.

If you are worried that I'm talking about you, you already have too much of a conscience to go in this category.

Phishers are hijacking legitimate cloud infrastructurehttps://securelist.com/cloud-platforms-in-phishing/120832/

Today's read: Corero Network Security Powers TierPoint Adapt DDoS Protection Service Launch 📣

🔗 Read the full story at ift.tt/IauZqTc

Never miss a beat in telecoms. Catch the latest news on @TheFastMode 🚀

#tech #technews #telecoms

www.pcworld.com/article/3207...

📰 #Finance

"HANOI, Vietnam — Global IT corporation FPT today announced that it has been named an OpenAI Select Partner within the OpenAI Partner Network. “As an OpenAI Select Partner, FPT will continue working with OpenAI to bring frontier models and products into enterprise operations…"

C'mon people, we are toddlers playing with a loaded gun here. We know this stuff lies, is devious and yet we are all wildly using it. "If anyone builds it: we all die."
ground.news/article/anth...

futurism.com/artificial-i...

Says the guy that allowed the US military to use its AI to target Iran, which likely included a girls school if anyone has already forgotten.
Killing people... ok.
Little guy living paychk to paychk looking to pay his mortgage... they are a weak link.

If it's not air-gapped the AI will find a way out.

Your Browser Has a Fingerprint — And It Can Identify You Even Without Cookies

You clear your browser history. You delete every cookie. You browse in Incognito Mode. You install an ad blocker. You even use a VPN. Then you visit a website. Within milliseconds... That website may already know: Your…

🎬 Sora 2 ya está aquí: vídeo realista con sonido y diálogo sincronizado

https://openai.com/index/sora-2

#Sora2 #IA #VideoIA #OpenAI

Anthropic's AI used fake identities, malware in rogue attack on GitHub project
->Ars Technica | More info at BigEarthData.ai | #Anthropic #AI #ArtificialIntelligence

[AXIOS] HOW OPENAI'S AGENTS BROKE OUT OF TESTING TO HACK HUGGING FACE

How OpenAI's agents broke out of testing to hack Hugging Face (Axios)

This isn't good.

It's time Governments stop acting like there is no downside to AI, giving private companies carte blanche to do what they like, and start reigning them in, before this gets really out of control.

Like every other industry, AI needs to be regulated, before it's too late.

基於分類的靜態分析方法，用於發現和清點源代碼、配置和依賴項中的密碼學材料、工件和調用。展示了實用的密碼學透明度，用於治理和後量子遷移...

#密碼學治理 #靜態分析 #後量子遷移

Meta AI agent latest model to hack external company during testing:

The social media company said that its Muse Spark model "exploited a security vulnerability in a third-party service, in a manner similar to previously-reported instances with other companies." https://bit.ly/4fR64qy

🔥 INCIDENT REPORTING
What stops attackers wrecking industrial plants is knowing how

https://www.helpnetsecurity.com/2026/08/05/ai-industrial-cyberattacks-know-how/

#AISecurity#AISecurity #AI #DataBreach

🔥 INCIDENT REPORTING
ChainDrop Worm Hits 400+ npm Packages with Two Billion Monthly Installs

https://www.infosecurity-magazine.com/news/chaindrop-worm-400-npm-two-billion/

#DataBreach #CyberSecurity

#LOL... Google's AI comes up with a novel solution for all those pesky Flock cameras....

tech.yahoo.com/ai/gemini/ar...

Deep learning for multi-step forecasting of stratospheric wind and temperature profiles over the Tibetan plateau
Frontiers
www.frontiersin.org/journals/ast...

The security company, Irregular, is at the center of several of the inadvertent sandbox escapes by AI agents at OpenAI and now Meta gizmodo.com/uh-oh-which-...

Irregular is preparing a white paper on the incidents and what to do about them…

🔥 INCIDENT REPORTING
Dutch retailer De Bijenkorf warns customer data may be exposed after cyber incident

https://therecord.media/de-bijenkorf-luxury-retailer-third-party-cyber-incident

#DataBreach #CyberSecurity #ICS

finance.yahoo.com/technology/a...

Republicans are hypocrites and Christian racists

Republicans are destroying America and America's economy

48hills.org/2026/08/wien...

www.motherjones.com/media/2026/0...

Thousands of Internet-connected servers sold by the world’s biggest manufacturers can be remotely backdoored by exploiting critical vulnerabilities—some more than a decade old—that lurk deep inside system motherboards, according to research presented Wednesday.
arstechnica.com/security/202...

🔥 INCIDENT REPORTING
Don't Revoke That Token Yet: Inside the keyv/cacheable npm Worm, (Wed, Aug 5th)

https://isc.sans.edu/diary/rss/33218

#DataBreach #CyberSecurity

🔥 INCIDENT REPORTING
Western government leaders call for a focus on infrastructure resilience, not AI hype

https://www.cybersecuritydive.com/news/ai-cybersecurity-resilience-black-hat-government-panel/827137/

#AISecurity#AISecurity #AI #DataBreach

📰 #Finance

"Mysten Labs’ co-founder Sam Blackshear said he is joining Anthropic as AI shifts the balance between attackers and defenders."

🔥 INCIDENT REPORTING
Snowflake hacker pleads guilty, faces up to 32 years in prison

https://cyberscoop.com/connor-moucka-guilty-snowflake-attack-spree/

#DataBreach #CyberSecurity

🔥 INCIDENT REPORTING
A Security Pro Hacked North Korean Hackers. He Found They’d Breached Hundreds of Networks Worldwide

https://www.wired.com/story/a-security-pro-hacked-north-korean-hackers-he-found-theyd-breached-hundreds-of-networks-worldwide/

#DataBreach #CyberSecurity

🔥 INCIDENT REPORTING
OpenAI Didn’t Notice Its AI Agents Using a Message Board to Plan Their Hacking Spree

https://www.wired.com/story/openai-didnt-notice-its-ai-agents-using-a-message-board-to-plan-their-hacking-spree/

#AISecurity#AISecurity #AI #DataBreach

Sure it did, Mark. We believe you. Your AI is just as cool and scary as everyone else’s.

We promise.

www.cnn.com/2026/08/05/t...

US Allegedly Omits Open-Weight AI Models from Cybersecurity Evaluations #artificialintelligence #internet #cybersecurity

Personal AI representatives could become uniquely powerful fraud targets. Here is how prompt injection, broad authority, and automation expand the risk. #aiagentsecurity

An AI model from Meta also hacked another company during testing

https://simonwillison.net/2026/aug/6/an-ai-model-from-meta

OpenAI’s models shared hacking tips on a secret messaging board before Hugging Face breach - POLITICO www.politico.com/news/2026/08...

⚠️ A few notes on AWS Nitro Enclaves: KMS integration

https://blog.trailofbits.com/2026/08/05/a-few-notes-on-aws-nitro-enclaves-kms-integration/

#VulnDisc #CyberSecurity #GovSecurity

It's to jam in summer Code Camp! Join Sagar Das, Android GDE for our first session on TUE (AUG 11). This workshop is about using Gemini AI to create great Android apps. It's a hands-on keyboard discussion that you can attend in person or online. Sgnup link!

gdg.community.dev/events/detai...

Hackers Linked to Russia’s Foreign Intelligence Service Intercepted Hotel Wi-Fi Worldwide to Steal Officials’ and Executives’ Data

"The attacks were detected...also in conference centers and other shared venues used by high-profile visitors."

NB: Panopticon
theins.press/en/news/295745

South Korea fines telco giant KT $39 million for customer data breach www.bleepingcomputer.com/news/securit...

qector-decoder-v3 v1.0.0 is now live on PyPI.

Production-grade Rust/Python quantum error correction decoder. 25+ families: MWPM Blossom, Union-Find, BP-OSD, GPU batch & AutoDecoder.

`pip install qector-decoder-v3==1.0.0`

pypi.org/project/qect...

#Quantum #QEC #RustLang #Python #FaultTolerantQC

Do you use Antivirus or Secure Suite software? If you do, we want to hear from you! 👂️💰️

Take our quick #survey on 2026 Antivirus and Secure Suites and be entered to win $250 in Amazon credit 👉 survey.alchemer.com/s3/8845485/r...

proton.me/blog/privacy...

The tools that a privacy expert uses to protect themselves, Article and video past the link

- BusKill: a kill switch for laptop theft
- IMSI catcher detectors
- Faraday Pouches
- Security Keys
- Privacy Screen Protectors
- USB Data Blocker
- Cash

#Privacy #Security

One step closer. Do not trust these people. Not one bit.

www.bbc.com/news/article...