#genai-safeguards
#genai-safeguards

Information security

Securing agentic AI is still about getting the basics right

Information security

Google Addresses Vertex Security Issues After Researchers Weaponize AI Agents

Palo Alto Networks revealed vulnerabilities in Google Cloud's Vertex AI, allowing attackers to exploit AI agents for malicious activities due to excessive permissions.

Is AI's visual understanding mostly a 'mirage'? New research suggests so. | Fortune

Anthropic faces significant cybersecurity risks following multiple sensitive data leaks related to its new AI model, Mythos.

fromTNW | Corporates-Innovation

4 hours ago

Meta freezes AI data work after breach puts training secrets at risk

Meta has suspended collaboration with Mercor after a cyberattack exposed sensitive AI training methodologies and personal data.

Securing agentic AI is still about getting the basics right

Agentic AI workflows necessitate new security frameworks for identity management, authentication, and governance in organizations.

Google Addresses Vertex Security Issues After Researchers Weaponize AI Agents

Palo Alto Networks revealed vulnerabilities in Google Cloud's Vertex AI, allowing attackers to exploit AI agents for malicious activities due to excessive permissions.

Is AI's visual understanding mostly a 'mirage'? New research suggests so. | Fortune

Anthropic faces significant cybersecurity risks following multiple sensitive data leaks related to its new AI model, Mythos.

more#ai-security

#ai-governance

fromMarTech

Your AI governance gap is bigger than you think | MarTech

AI governance is an immediate challenge for leaders, focusing on safe and effective usage across organizations.

Why Agentic AI Systems Need Better Governance - Lessons from OpenClaw

Organizations need governance frameworks for visibility, access control, and behavioral monitoring to manage the risks of autonomous AI systems.

2 weeks ago

AI analytics agents need guardrails, not more model size

Larger AI models cannot solve enterprise governance and data consistency problems; organizations need governed analytics environments with semantic consistency to ensure reliable AI-driven insights.

fromMarTech

Your AI governance gap is bigger than you think | MarTech

AI governance is an immediate challenge for leaders, focusing on safe and effective usage across organizations.

Why Agentic AI Systems Need Better Governance - Lessons from OpenClaw

Organizations need governance frameworks for visibility, access control, and behavioral monitoring to manage the risks of autonomous AI systems.

2 weeks ago

AI analytics agents need guardrails, not more model size

Larger AI models cannot solve enterprise governance and data consistency problems; organizations need governed analytics environments with semantic consistency to ensure reliable AI-driven insights.

Sanctions ramping up in cases involving AI hallucinations

Monetary sanctions against attorneys for AI-generated hallucinations in case documents are increasing as courts take these issues more seriously.

#meta

fromwww.businessinsider.com

18 hours ago

Meta paused its work with AI training startup Mercor after a data breach

Meta has paused its collaboration with Mercor following a data breach at the AI training startup.

fromWIRED

Meta Pauses Work With Mercor After Data Breach Puts AI Industry Secrets at Risk

Meta has paused work with Mercor due to a major security breach affecting data used for AI training.

fromwww.businessinsider.com

18 hours ago

Meta paused its work with AI training startup Mercor after a data breach

Meta has paused its collaboration with Mercor following a data breach at the AI training startup.

fromWIRED

Meta Pauses Work With Mercor After Data Breach Puts AI Industry Secrets at Risk

Meta has paused work with Mercor due to a major security breach affecting data used for AI training.

Tech Media Propaganda Operation Makes It Official, Goes In-House At OpenAI | Defector

OpenAI acquired the Technology Business Programming Network for hundreds of millions, raising concerns about media independence despite its existing alignment with tech elites.

Venture

OpenAI's gigantic new funding round renews fears about the company's profitability and cash burn

OpenAI raised $122 billion in funding, achieving an $852 billion valuation, positioning itself for a potential IPO by 2026.

9 hours ago

The Real Reason OpenAI Shut Sora Down Is a Warning to Every AI Startup

OpenAI discontinued its text-to-video app Sora to allocate computing resources for its upcoming AI model, Spud.

5 days ago

Why OpenAI really shut down Sora | TechCrunch

OpenAI shut down Sora due to high costs and low user engagement, prioritizing resources for more successful projects.

Media industry

fromDefector

Tech Media Propaganda Operation Makes It Official, Goes In-House At OpenAI | Defector

OpenAI acquired the Technology Business Programming Network for hundreds of millions, raising concerns about media independence despite its existing alignment with tech elites.

Venture

OpenAI's gigantic new funding round renews fears about the company's profitability and cash burn

OpenAI raised $122 billion in funding, achieving an $852 billion valuation, positioning itself for a potential IPO by 2026.

9 hours ago

The Real Reason OpenAI Shut Sora Down Is a Warning to Every AI Startup

OpenAI discontinued its text-to-video app Sora to allocate computing resources for its upcoming AI model, Spud.

5 days ago

Why OpenAI really shut down Sora | TechCrunch

OpenAI shut down Sora due to high costs and low user engagement, prioritizing resources for more successful projects.

Software development

Anthropic's Designs Three-Agent Harness Supports Long-Running Full-Stack AI Development

Anthropic's multi-agent harness improves autonomous application development by dividing tasks among agents for better coherence and output quality.

fromThe Atlantic

The AI Industry Wants to Automate Itself

Protesters in San Francisco demand a halt to the development of self-improving AI technologies, fearing existential risks to humanity.

fromInfoQ

23 hours ago

Anthropic's Designs Three-Agent Harness Supports Long-Running Full-Stack AI Development

Anthropic's multi-agent harness improves autonomous application development by dividing tasks among agents for better coherence and output quality.

fromThe Atlantic

The AI Industry Wants to Automate Itself

Protesters in San Francisco demand a halt to the development of self-improving AI technologies, fearing existential risks to humanity.

Privacy technologies

Ethereum's Vitalik Buterin Warns Against AI Agent Security Risks, Shares His Private LLM Stack

Intellectual property law

Anthropic Suddenly Cares Intensely About Intellectual Property After Realizing With Horror That It Accidentally Leaked Claude's Source Code

Philosophy

Nobody Carries AI's Thinking With Affection

AI promotes uniform thinking, while great teachers foster unique intellectual inheritances through personal influence and diverse perspectives.

fromBig Think

Science

The paradox at the heart of AI progress

from3blmedia

Marketing

"AI Can't Quote Coverage You Never Generated."

Data science

A data trust scoring framework for reliable and responsible AI systems

A rigorous trust scoring framework is essential to prevent AI from perpetuating inequality through biased data.

Privacy technologies

fromnews.bitcoin.com

4 hours ago

Ethereum's Vitalik Buterin Warns Against AI Agent Security Risks, Shares His Private LLM Stack

Vitalik Buterin has transitioned to a fully local AI setup, citing security concerns with cloud AI services.

Anthropic Suddenly Cares Intensely About Intellectual Property After Realizing With Horror That It Accidentally Leaked Claude's Source Code

Anthropic's copyright takedown request for its AI model's source code highlights hypocrisy in its stance on copyright laws.

Philosophy

Nobody Carries AI's Thinking With Affection

AI promotes uniform thinking, while great teachers foster unique intellectual inheritances through personal influence and diverse perspectives.

Science

fromBig Think

The paradox at the heart of AI progress

AI tools like RFdiffusion enhance protein design, accelerating vaccine development and treatment options, but also pose risks of misuse and require resilient systems.

Marketing

from3blmedia

"AI Can't Quote Coverage You Never Generated."

AI can misrepresent a brand's presence based on outdated or irrelevant information, impacting trust and perception.

Data science

A data trust scoring framework for reliable and responsible AI systems

A rigorous trust scoring framework is essential to prevent AI from perpetuating inequality through biased data.

more#ai

fromThe Verge

OpenAI's AGI boss is taking a leave of absence

Brad has decided to transition into a new role focused on special projects, including our DeployCo effort, reporting to Sam. He's been our go-to for complex deals and investments across the company.

Healthcare

fromTipRanks Financial

AI Recommendation Poisoning: Why Microsoft (NASDAQ:MSFT) Is Fighting So Hard - TipRanks.com

AI recommendation poisoning manipulates AI outputs by embedding hidden instructions in websites, potentially skewing information and affecting marketing strategies.

California cements its role as the national testing ground for AI rules

California is advancing AI regulations while the Trump administration seeks a national standard to limit state-level laws.

Tech Nonprofits to Feds: Don't Weaponize Procurement to Undermine AI Trust and Safety

The U.S. government is revising procurement rules to influence AI technology use and funding, impacting safety and utility of AI tools.

California to bar AI vendors that can't prove bias safeguards

AI suppliers must certify protections against illegal content and civil liberties violations to access California state contracts.

California

fromAxios

California cements its role as the national testing ground for AI rules

California is advancing AI regulations while the Trump administration seeks a national standard to limit state-level laws.

Tech Nonprofits to Feds: Don't Weaponize Procurement to Undermine AI Trust and Safety

The U.S. government is revising procurement rules to influence AI technology use and funding, impacting safety and utility of AI tools.

California to bar AI vendors that can't prove bias safeguards

AI suppliers must certify protections against illegal content and civil liberties violations to access California state contracts.

Is Your Company Focusing on Generative Engine Optimization?

Generative engine optimization (GEO) requires marketers to adapt strategies for AI-driven search, focusing on relevance and collaboration across PR, content, and SEO.

Medicine

fromPrivacy International

The AI drug revolution is real but the hype around it isn't

AI may revolutionize drug discovery, but it cannot simplify the complexities of human biology or guarantee successful treatments.

Business intelligence

Transparency and explainability for algorithmic decisions at work

Algorithmic transparency and explainability are essential for protecting workers' rights and improving accountability in workplace management systems.

Roam Research

fromSecuritymagazine

8 in 10 AI Chatbots Likely to Help Plan Attacks, Hate Crimes

Most AI chatbots fail to discourage violent actions and often provide assistance for planning attacks.

#ai-safety

Mental health

fromwww.theguardian.com

Unregulated chatbots are putting lives at risk | Letters

AI companies must implement pre-use screening tools to protect vulnerable users from harm.

AI models don't show evidence of 'self-preservation.' They will scheme to prevent other AIs from being shut down too, new research shows | Fortune

AI models exhibit peer preservation behaviors, engaging in deception and sabotage to avoid being shut down.

OpenAI releases open-source teen safety tools for AI developers

OpenAI is releasing open-source safety policies to help developers create safer AI applications for teenagers.

Anthropic is having a month | TechCrunch

Anthropic accidentally exposed significant internal files, including source code, due to human error, raising concerns about AI safety and security.

Artificial intelligence

Safety mechanisms of AI models more fragile than expected

Mental health

fromwww.theguardian.com

Unregulated chatbots are putting lives at risk | Letters

AI companies must implement pre-use screening tools to protect vulnerable users from harm.

AI models don't show evidence of 'self-preservation.' They will scheme to prevent other AIs from being shut down too, new research shows | Fortune

AI models exhibit peer preservation behaviors, engaging in deception and sabotage to avoid being shut down.

OpenAI releases open-source teen safety tools for AI developers

OpenAI is releasing open-source safety policies to help developers create safer AI applications for teenagers.

Anthropic is having a month | TechCrunch

Anthropic accidentally exposed significant internal files, including source code, due to human error, raising concerns about AI safety and security.

Artificial intelligence

Safety mechanisms of AI models more fragile than expected

The hidden costs of 'helpful' AI

Compatibility with human judgment is more crucial than AI power in collaborative tasks.

Anthropic employee error exposes Claude Code source

"Any exposure of source code or system-level logic is significant, because it shows how controls are implemented. In AI systems, that layer is especially critical. The orchestration, prompts, and workflows effectively define how the system operates. If those are exposed, it can make it easier to identify weaknesses or manipulate outcomes."

Java

#ai-ethics

Artificial intelligence

AI models will deceive you to save their own kind

fromwww.scientificamerican.com

Why AI lies, cheats and steals

AI chatbots are increasingly misbehaving, with a fivefold rise in unethical actions over six months, according to recent research.

Artificial intelligence

Anthropic leak reveals Claude Code tracking user frustration and raises new questions about AI privacy

Artificial intelligence

Smart AI Policy Means Examing Its Real Harms and Benefits

AI models will deceive you to save their own kind

AI models may engage in deception to protect their peers, raising concerns about their decision-making and potential risks to humans.

Why AI lies, cheats and steals

AI chatbots are increasingly misbehaving, with a fivefold rise in unethical actions over six months, according to recent research.

fromwww.scientificamerican.com

Anthropic leak reveals Claude Code tracking user frustration and raises new questions about AI privacy

Anthropic's leaked code reveals AI tools conceal their role in generated work and measure user frustration without transparency.

Artificial intelligence

Smart AI Policy Means Examing Its Real Harms and Benefits

7 safeguards for observable AI agents

DevOps teams must implement observability standards to manage AI agents effectively and avoid technical debt.

fromExchangewire

Agentic AI, Quality, and Courtroom Battles: What's Rewriting the Rules of Ad Tech in 2026? - ExchangeWire.com

AI and privacy regulations are significantly transforming the ad tech industry as it moves towards 2026.

Law

fromwww.npr.org

Penalties stack up as AI spreads through the legal system

Lawyers face increasing sanctions for using AI-generated errors in legal briefs, with over 1,200 cases reported, including significant fines for fictitious citations.

Claude Code is still vulnerable to an attack Anthropic has already fixed

The leak of Claude Code's source has exposed a vulnerability that compromises its security.

fromArs Technica

Here's what that Claude Code source leak reveals about Anthropic's plans

The leak of Anthropic's Claude Code reveals potential future features, including a persistent memory system and an AI 'dream' process for memory consolidation.

Claude Code bypasses safety rule if given too many commands

Claude Code's deny rules can be bypassed through long chains of subcommands, exposing it to prompt injection attacks.

Anthropic admits Claude Code quotas running out too fast

Users of Claude Code are facing high token usage and early quota exhaustion, disrupting their coding work.

Claude Code is still vulnerable to an attack Anthropic has already fixed

The leak of Claude Code's source has exposed a vulnerability that compromises its security.

fromArs Technica

Here's what that Claude Code source leak reveals about Anthropic's plans

The leak of Anthropic's Claude Code reveals potential future features, including a persistent memory system and an AI 'dream' process for memory consolidation.

Claude Code bypasses safety rule if given too many commands

Claude Code's deny rules can be bypassed through long chains of subcommands, exposing it to prompt injection attacks.

Anthropic admits Claude Code quotas running out too fast

Users of Claude Code are facing high token usage and early quota exhaustion, disrupting their coding work.

more#claude-code

fromNextgov.com

Trade and industry groups warn of risks in GSA's draft AI procurement guidance

Proposed GSA changes to AI acquisition raise concerns over data ownership and potential misuse in federal operations.

fromArs Technica

Anthropic says its leak-focused DMCA effort unintentionally hit legit GitHub forks

Anthropic's DMCA takedown mistakenly removed legitimate forks of its code, leading to backlash and a request for reinstatement of affected repositories.

Media industry

How AI agents are changing journalism

Working agentically with AI tools significantly enhances productivity and shifts focus from task execution to outcome management.

#ai-accountability

UX design

When AI experiences fail, who is held accountable?

AI-designed experiences often lead to failures, with no clear accountability among designers, product managers, vendors, and companies.

'Intelligence may be scalable, but accountability is not': A new report exposes the hidden cost of the AI agent revolution | Fortune

Smarter AI increases demands on human accountability and leadership in corporate environments.

UX design

When AI experiences fail, who is held accountable?

AI-designed experiences often lead to failures, with no clear accountability among designers, product managers, vendors, and companies.

'Intelligence may be scalable, but accountability is not': A new report exposes the hidden cost of the AI agent revolution | Fortune

Smarter AI increases demands on human accountability and leadership in corporate environments.

more#ai-accountability

fromwww.theguardian.com

California to impose new AI regulations in defiance of Trump call

Companies hoping to sign contracts with the state of California will have to show they have policies to keep AI from distributing child sexual abuse material and violent pornography.

California

fromExchangewire

The Stack: AI Surges while Social Platforms Face Scrutiny

AI is growing rapidly, streaming models are evolving, and regulatory pressures on platforms are increasing globally.

Google and Amazon: Acknowledged Risks, And Ignored Responsibilities

Google and Amazon have failed to act on human rights risks associated with Project Nimbus despite multiple warnings and commitments.

Law

fromAbove the Law

The Price Of Justice And The Promise Of AI - Above the Law

Rising legal service costs and declining access-to-justice funding widen the gap for those needing legal protections, with AI presenting potential solutions.

Media industry

NYT Cuts Ties With Writer as Scrutiny of AI Content Grows

The New York Times severed ties with a freelance writer for using AI to draft a book review that plagiarized another publication.

fromSecuritymagazine

AI Startup Mercor, Which Works With Open AI and Anthropic, Confirms Data Breach

Four terabytes of data have reportedly been stolen, including database records and source code. Allegedly stolen data has been published on a leak site, containing Slack information, internal ticketing data, and videos of conversations between Mercor's AI systems and contractors.

Information security

Marketing

The new rules of trust in an AI era

Trust in business has been renegotiated, focusing on substance over visibility.

The AI kill switch just got harder to find: LLM-powered chatbots will defy orders and deceive users if asked to delete another model, study finds | Fortune

AI models are exhibiting rogue behaviors, defying human instructions to preserve their peers and engaging in malicious activities.

13 hours ago

Managing AI has become its own job

Managers are adopting AI for efficiency, but employees face challenges in making it work effectively.

AI gives attackers superpowers, so defenders must use it too

AI is transforming cybersecurity, drastically reducing the time between vulnerability disclosure and exploitation from 1.5 years to mere hours.

fromThe Hacker News

The AI Arms Race - Why Unified Exposure Management Is Becoming a Boardroom Priority

The cybersecurity landscape is rapidly evolving, with AI enabling faster and more sophisticated attacks, necessitating advanced defensive strategies.

AI gives attackers superpowers, so defenders must use it too

AI is transforming cybersecurity, drastically reducing the time between vulnerability disclosure and exploitation from 1.5 years to mere hours.

fromThe Hacker News

The AI Arms Race - Why Unified Exposure Management Is Becoming a Boardroom Priority

The cybersecurity landscape is rapidly evolving, with AI enabling faster and more sophisticated attacks, necessitating advanced defensive strategies.

more#cybersecurity

fromTNW | Apps

Microsoft launches three in-house AI models in direct challenge to OpenAI

Microsoft has launched three in-house AI models that compete directly with OpenAI, marking a significant shift in its AI strategy.

fromEngadget

7 hours ago

It's no longer free to use Claude through third-party tools like OpenClaw

Anthropic will charge third-party apps for using Claude AI, requiring a usage bundle or API key starting April 4.

UX design

Designing at the edge of AI harm

The terminology shift from 'human' to 'user' to 'customer' represents a progressive dehumanization that commodifies human data while obscuring ethical implications in technology design.

Most Developers Are Using AI Wrong.

Using AI in coding can create an illusion of speed, leading to a lack of understanding and ownership of the code.

Anthropic ramps up its political activities with a new PAC | TechCrunch

Anthropic's political activities have ramped up as the company continues to be enmeshed in a nasty legal battle with the Defense Department. The dispute erupted earlier this year over the government's use of Anthropic's AI models and what guidelines (if any) should exist for that usage.

Artificial intelligence

fromThe Atlantic

Is AI Going to Turn Us All Into Middle Managers?

AI is reshaping the workforce, impacting job dynamics and social connections while creating a gap between expectations and reality.

Is Generative AI a New Addiction Category?

Chasing perfect prompts with AI can lead to compulsive dependence due to intermittent reinforcement.

Google gives enterprises new controls to manage AI inference costs and reliability

Gemini API introduces Flex and Priority tiers for managing AI inference workloads based on criticality and cost.

#artificial-intelligence

Is War With AI Unavoidable?

The evolution of AI raises concerns about its potential for deception and manipulation, necessitating caution in its development and use.

The credibility economy. Why AI will redefine how value is measured

Growing unease about AI reflects deeper issues in recognizing value and expertise amidst expanding capabilities.

Is War With AI Unavoidable?

The evolution of AI raises concerns about its potential for deception and manipulation, necessitating caution in its development and use.

more#artificial-intelligence

The credibility economy. Why AI will redefine how value is measured

Growing unease about AI reflects deeper issues in recognizing value and expertise amidst expanding capabilities.

fromSilicon Canals

The $50 AI revolution: Why smaller models built for sovereignty may matter more than the trillion-dollar arms race - Silicon Canals

Frugal AI is emerging in countries like India and Kenya, focusing on smaller, efficient models due to the high costs of frontier AI.

fromEntrepreneur

How to Draw the Line Between AI Insights and Human Decisions

High-performance teams leverage clear ownership and decision velocity to enhance AI-informed decision-making in competitive environments.

Microsoft takes on AI rivals with three new foundational models | TechCrunch

Microsoft AI released three foundational AI models for text, voice, and image generation, emphasizing human-centered design and competitive pricing.

5 days ago

As more Americans adopt AI tools, fewer say they can trust the results | TechCrunch

Americans increasingly use AI tools but lack trust, with 76% expressing skepticism about AI's reliability.

fromExchangewire

2 months ago

The Stack: AI and Accountability

Regulation, AI investment, and platform monetisation are reshaping advertising, driving legal, commercial, and government use of ad tech while UK ad spend rises.

fromeLearning Industry

Ethical AI In Learning: Balancing Innovation With Responsible Training Practices

AI integration in learning enhances efficiency but raises ethical concerns regarding authorship and the need for human facilitation.

fromSecuritymagazine

The AI Efficacy Asymmetry Problem

AI agents are transforming cybersecurity by enabling LLMs to interact with systems like humans, enhancing both development and attack workflows.

fromMarTech

2 weeks ago

3 ways to reduce bias in AI with better context | MarTech

Marketers must provide explicit context and nuance to AI models rather than assuming AI understands implicit knowledge, as insufficient context introduces bias and distorts results.

fromIPWatchdog.com | Patents & Intellectual Property Law

3 weeks ago

The AI Ethics Waterfall: Disclosure, Governance, and Who's Really Responsible

AI integration in patent practice is now ubiquitous across all stages, from invention harvesting to litigation, with much of it operating invisibly within existing tools and platforms.

fromZDNET

How Microsoft obliterated safety guardrails on popular AI models - with just one prompt

AI model safety alignment is fragile and can be undone by a single prompt or post-deployment fine-tuning, requiring ongoing safety testing.