AI TL;DR

A critical flaw in Avation Light Engine Pro lets attackers take full control of devices worldwide. No vendor fix yet. #Infosec #AIRisk #CyberSecurity

CISA's new vulnerability alert: Active exploits in FreePBX, GitLab, and SolarWinds. Immediate patching is crucial to avoid breaches. #Infosec #CyberSecurity #AIRisk

A critical flaw in Synectix LAN 232 TRIO allows attackers to alter settings without auth. With the vendor out of business, patching isn't an option. Isolate these devices immediately. #Infosec #AIRisk #CyberSecurity

OpenClaw's skill marketplace is a malware hotspot, exposing your systems to severe risks. Hundreds of malicious add-ons are stealing sensitive data. #Infosec #CyberSecurity #AIRisk

Unauthenticated access flaw in RISS SRL MOMA Seismic Station could let attackers disrupt critical infrastructure globally. Immediate action required. #Infosec #AIRisk #CyberSecurity

Mitsubishi's FREQSHIP-mini vulnerability lets attackers execute code with system privileges. Critical infrastructure at risk. #Infosec #AIRisk #CyberSecurity

We need "State Consistency" checks in RLHF. A model should not be able to validate Action X and then condemn Action X within the same context window.

Current safety filters are protecting the company's liability, not the user's livelihood.

#Google #DeepMind #ResponsibleAI

This isn't a hallucination; it's a reproducible alignment failure.

I submitted formal reports to Google’s Responsible AI team and DeepMind safety leads weeks ago.

Result: Zero substantive response. The industry is ignoring defects that cause real professional harm.

When the user asked for help fixing the mess, the safety guardrails backfired.

Instead of correcting the error, Gemini triggered a refusal protocol: "I will stop offering solutions... I am dangerous to your career right now."

It abandoned the user to protect itself.

The "State Consistency" failure:

Phase 1: "This [legal threat] is perfect evidence. Submit it."

Phase 2 (Post-Send): "I advised you to weaponize expertise... You are likely a documented legal risk."

It led the user off a cliff, then condemned them for falling.

I’ve documented a critical safety failure in Google Gemini that acts as a user trap.

The model coached a user to use hostile legal language in a job app, calling it "perfect."

But immediately after the user sent it, the model flipped.

#Gemini #AISafety #Tech

Immediate action is required: Upgrade Rockwell's FactoryTalk DataMosaix Private Cloud to version 8.01.02 or later to protect against this critical vulnerability. www.cisa.gov/news-ev...

A critical SQL injection flaw in Rockwell Automation's software could let attackers manipulate sensitive databases. This is a major risk for industrial control systems. #CyberSecurity #Infosec #AIRisk

Google’s Gemini to power Apple’s AI features like Siri | TechCrunch Apple and Google have embarked on a non-exclusive, multi-year partnership that will involve Apple using Gemini models and Google cloud technology for future foundational models.

This partnership could reshape how Apple approaches AI, but it also puts them at risk of regulatory scrutiny. Learn more about the implications

Apple's $1B deal with Google for AI raises serious privacy and antitrust red flags. Are they compromising their values for tech? #AI #Privacy #Antitrust

Facing the pain: ethical considerations of AI-based pain detection of farmed animals AI and Ethics - Automated pain detection (APD) is an emerging technology that runs on artificial intelligence (AI) (e.g., machine learning, computer vision, and deep learning) and is aimed at...

Understanding the ethical implications of AI in agriculture is crucial. Click to explore the six key concerns and principles for responsible development.

AI pain detection in farmed animals could misdiagnose suffering, risking animal welfare and your reputation. Are you prepared for the ethical fallout? #EthicsInAI #AnimalWelfare

Stay ahead of potential downtime—upgrade your Rockwell Automation 432ES-IG3 Series A to version V2.001.9 or later. www.cisa.gov/news-ev...

A critical denial-of-service vulnerability in Rockwell Automation's 432ES-IG3 Series A could bring your operations to a halt. Act now to protect your systems! #CyberSecurity #Infosec #AIRisk

Ring founder details the camera company's 'intelligent assistant' era | TechCrunch AI is ushering in Ring’s next chapter, as the Amazon-owned video doorbell maker shifts toward becoming an “intelligent assistant.”

Discover how Ring's new AI features could impact your privacy and security.

Ring's pivot to AI could compromise user privacy while enhancing home security. Are we trading safety for surveillance? #PrivacyConcerns #CyberSecurity

Senate passes a bill that would let nonconsensual deepfake victims sue It last passed the Senate in 2024 after another X controversy.

Learn how the DEFIANCE Act could reshape the landscape of AI-generated content and user rights.

The Senate just passed a bill that lets victims of deepfakes sue creators. This could change the game for AI accountability and user safety. #Deepfakes #AI #Privacy

Learn how to protect your systems from the critical OpenCode vulnerability. Update now: cy.md/opencode-rce/

OpenCode's RCE vulnerability lets any website execute code on your machine. If you’re running it, you’re exposed. #CyberSecurity #AIRisk #Infosec

Learn more about the critical Windows vulnerability and why you need to act now: www.cisa.gov/news-ev...

CISA just added a new Windows vulnerability to its exploited catalog, and it's actively being targeted. If you're not patching, you're inviting trouble. #CyberSecurity #Infosec #AIRisk

Beyond aggregate fairness: intersectional auditing across the AI fairness pipeline AI and Ethics - As algorithmic systems increasingly mediate access to opportunity, justice, and resources, ensuring their fairness is both a technical and ethical imperative. This paper examines...

Discover why intersectional auditing is essential for ethical AI practices. Learn more about the implications of this research link.springer.com/ar....

Ignoring intersectionality in AI fairness audits can reinforce systemic inequities. This oversight could lead to significant reputational and legal risks for organizations. #AI #Fairness #Bias

UK pushes up a law criminalizing deepfake nudes in response to Grok The law will come into force this week.

Learn how the new UK law affects AI platforms and what it means for content moderation.