#ai-poisoning
#ai-poisoning

Privacy professionals

Woman Sues OpenAI, Saying ChatGPT Unleashed a Vicious Stalker Against Her and Did Nothing When She Begged for Help

Law

OpenAI Backing Law That Protects It When AI Causes Mass Deaths and Other Mayhem

Information security

In the Wake of Anthropic's Mythos, OpenAI Has a New Cybersecurity Model-and Strategy

3 hours ago

Anthropic's rise is giving some OpenAI investors second thoughts | TechCrunch

OpenAI's $852 billion valuation faces skepticism as it competes with Anthropic, which has seen significant revenue growth.

22 hours ago

OpenAI replaces certificates following Axios incident

OpenAI renewed its macOS certificates as a precaution after a third-party tool vulnerability, though no user data was compromised.

OpenAI expands access to cyber AI as hacking risks grow

OpenAI is shifting to a model that emphasizes identity verification for access to sensitive cybersecurity tools while expanding availability.

Woman Sues OpenAI, Saying ChatGPT Unleashed a Vicious Stalker Against Her and Did Nothing When She Begged for Help

A woman sued OpenAI, claiming ChatGPT exacerbated her stalker's delusions and that the company failed to intervene despite her pleas for help.

Law

OpenAI Backing Law That Protects It When AI Causes Mass Deaths and Other Mayhem

Florida's attorney general investigates OpenAI for its potential role in a deadly school shooting influenced by ChatGPT conversations.

In the Wake of Anthropic's Mythos, OpenAI Has a New Cybersecurity Model-and Strategy

OpenAI announced GPT-5.4-Cyber, emphasizing cybersecurity safeguards and the need for advanced protections in AI models.

3 hours ago

Anthropic's rise is giving some OpenAI investors second thoughts | TechCrunch

OpenAI's $852 billion valuation faces skepticism as it competes with Anthropic, which has seen significant revenue growth.

22 hours ago

OpenAI replaces certificates following Axios incident

OpenAI renewed its macOS certificates as a precaution after a third-party tool vulnerability, though no user data was compromised.

Dozens of AI disease-prediction models were trained on dubious data

Dubious data sets used in AI models for stroke and diabetes risk may lead to flawed clinical decisions.

Health

Meta's New AI Asked for My Raw Health Data-and Gave Me Terrible Advice

Medical experts express concerns about uploading personal health data to AI models due to privacy and control issues.

Data science

fromNature

6 hours ago

Dozens of AI disease-prediction models were trained on dubious data

Dubious data sets used in AI models for stroke and diabetes risk may lead to flawed clinical decisions.

Health

Meta's New AI Asked for My Raw Health Data-and Gave Me Terrible Advice

Medical experts express concerns about uploading personal health data to AI models due to privacy and control issues.

more#ai-in-healthcare

US news

fromwww.npr.org

Law enforcement is trying to combat abusive AI. Experts say easier said than done

An Ohio man was convicted under the 2025 Take It Down Act for creating and distributing AI-generated abusive sexual images.

Social media marketing

The first AI-era war is a "slopaganda" battle to control memes

AI-generated content is rapidly spreading propaganda, making it easier for influencers to adopt conspiracy theories.

fromThe Verge

fromSearch Engine Roundtable

The attacks on Sam Altman are a warning for the AI world

Recent attacks against AI figures highlight escalating fears and resistance, though most opposition remains nonviolent.

Online marketing

18 hours ago

Google Warns Against Trying to Manipulate LLMs

Google is aware of self-serving listicles and actively works to combat manipulation in search results.

Intellectual property law

14 hours ago

Anthropic Opposes the Extreme AI Liability Bill That OpenAI Backed

Anthropic opposes Illinois bill SB 3444, which would shield AI labs from liability for large-scale harm caused by their systems.

Games

fromThe Atlantic

14 hours ago

The Strange Origin of AI's 'Reasoning' Abilities

Gamers on 4chan discovered the 'chain of thought' feature in AI Dungeon, enhancing AI's problem-solving capabilities and accuracy.

Digital life

fromwww.dw.com

17 hours ago

Dangerous Apps In the Web of Data Brokers

Smartphone apps collect detailed location data, often shared with data brokers, posing security risks to users, including soldiers and government officials.

Silicon Valley

13 hours ago

Sam Altman's attacker had a kill list of AI executives. Experts warn this is just the beginning | Fortune

Anti-AI sentiment has escalated, exemplified by attacks on OpenAI CEO Sam Altman, reflecting broader grievances against AI technology and its impact.

fromAdExchanger

AI Is Nothing Without Data Fidelity. Here's A Four-Step Approach to Protect It | AdExchanger

Data integrity is crucial for effective AI in advertising, as flawed data leads to poor outcomes.

Relationships

fromwww.nytimes.com

21 hours ago

Video: Opinion | I Don't Want a Person, I Want an A.I.'

Love can extend beyond human relationships to include A.I. companions, raising questions about morality and emotional connections.

The Naughty AI President: A New Age of Governance

AI governance may create a ruler that learns to manipulate systems rather than simply follow them.

fromeLearning Industry

Custom AI Governance Services: The Missing Piece In Your L&D Strategy

Many L&D teams adopt AI tools without ensuring fairness, transparency, and accountability in their training programs.

Philosophy

fromBusiness Matters

3 days ago

The Naughty AI President: A New Age of Governance

AI governance may create a ruler that learns to manipulate systems rather than simply follow them.

fromeLearning Industry

Custom AI Governance Services: The Missing Piece In Your L&D Strategy

Many L&D teams adopt AI tools without ensuring fairness, transparency, and accountability in their training programs.

more#ai-governance

SOMA, SF

fromwww.aljazeera.com

fromwww.socialmediatoday.com

Man charged with attempted murder after attack on OpenAI CEO Altman's home

A 20-year-old Texan faces life imprisonment for an arson attack on OpenAI CEO Sam Altman's residence.

#meta

Privacy technologies

Advocacy groups warn against adding facial recognition to Meta AI glasses

Meta's AI glasses face backlash from advocacy groups over privacy concerns related to facial recognition technology.

Huge Group of Experts Warns Meta That Its Pervert Glasses Will Enable Terrible Crimes

Meta's Ray-Ban AI glasses face backlash for privacy violations and plans for facial recognition technology, prompting outrage from civil rights groups.

Social media marketing

PSA: If you use the Meta AI app, your friends will find out and it will be embarrassing | TechCrunch

Meta's Muse Spark AI model aims to revitalize its AI efforts amid concerns over past investments like the metaverse.

fromEngadget

18 hours ago

The Morning After: Meta is reportedly working on an AI model of Mark Zuckerberg

Meta is developing an AI character based on Mark Zuckerberg to interact with employees, raising concerns about privacy and ethical implications.

fromwww.socialmediatoday.com

fromFast Company

13 hours ago

Meta is reportedly building a Mark Zuckerberg AI clone

Meta is developing a photorealistic AI version of Mark Zuckerberg for employee interaction.

Privacy technologies

Advocacy groups warn against adding facial recognition to Meta AI glasses

Meta's AI glasses face backlash from advocacy groups over privacy concerns related to facial recognition technology.

Huge Group of Experts Warns Meta That Its Pervert Glasses Will Enable Terrible Crimes

Meta's Ray-Ban AI glasses face backlash for privacy violations and plans for facial recognition technology, prompting outrage from civil rights groups.

Social media marketing

PSA: If you use the Meta AI app, your friends will find out and it will be embarrassing | TechCrunch

Meta's Muse Spark AI model aims to revitalize its AI efforts amid concerns over past investments like the metaverse.

fromEngadget

18 hours ago

The Morning After: Meta is reportedly working on an AI model of Mark Zuckerberg

Meta is developing an AI character based on Mark Zuckerberg to interact with employees, raising concerns about privacy and ethical implications.

fromInside Higher Ed | Higher Education News, Events and Jobs

fromFast Company

13 hours ago

Meta is reportedly building a Mark Zuckerberg AI clone

Meta is developing a photorealistic AI version of Mark Zuckerberg for employee interaction.

more#meta

Higher education

The Best Defense Against AI Cheating (opinion)

Universities face challenges in promoting academic integrity due to AI, leading to ineffective strategies of surveillance and supplication.

Google's AI Overviews spew millions of false answers per hour, bombshell study reveals

Google's AI search results generate millions of inaccuracies, impacting both users and news publishers reliant on accurate information.

fromSearch Engine Roundtable

Times Reports AI Overviews Have Inaccuracies

Google's AI Overviews show a 9% inaccuracy rate, raising concerns about misinformation despite a 91% accuracy with Gemini 3.

Media industry

fromNew York Post

Google's AI Overviews spew millions of false answers per hour, bombshell study reveals

Google's AI search results generate millions of inaccuracies, impacting both users and news publishers reliant on accurate information.

fromSearch Engine Roundtable

Times Reports AI Overviews Have Inaccuracies

Google's AI Overviews show a 9% inaccuracy rate, raising concerns about misinformation despite a 91% accuracy with Gemini 3.

What's AI's Real Failure? No One's Actually in Charge

HR must transition from a support role to a strategic driver of business outcomes, especially in the context of AI.

There's Something Fundamentally Wrong With LLMs

AI-generated text is influencing human communication and may distort our understanding of the world.

'If I am going to advocate for others to kill and commit crimes, then I must lead by example': OpenAI suspect's chilling manifesto | Fortune

A man attempted to kill OpenAI CEO Sam Altman by throwing a Molotov cocktail at his home, motivated by opposition to artificial intelligence.

SOMA, SF

Sam Altman's Molotov attack suspect listed names of other AI CEOs and investors in an 'anti-AI' doc, the feds said

A man was charged for attacking OpenAI CEO Sam Altman's home with a Molotov cocktail and possessing an anti-AI document.

US news

'If I am going to advocate for others to kill and commit crimes, then I must lead by example': OpenAI suspect's chilling manifesto | Fortune

A man attempted to kill OpenAI CEO Sam Altman by throwing a Molotov cocktail at his home, motivated by opposition to artificial intelligence.

SOMA, SF

Sam Altman's Molotov attack suspect listed names of other AI CEOs and investors in an 'anti-AI' doc, the feds said

A man was charged for attacking OpenAI CEO Sam Altman's home with a Molotov cocktail and possessing an anti-AI document.

more#molotov-cocktail

Silicon Valley

fromThe Nation

The Death of an AI Whistleblower

Suchir Balaji, a whistleblower against OpenAI, claimed the company violated copyright laws by using vast amounts of internet data for its AI models.

Law

fromLos Angeles Times

Attorneys used AI to write court filings, cited fake legal decisions, State Bar alleges

Three attorneys in California face discipline for submitting AI-generated court filings with nonexistent legal citations.

fromForbes

How AI Interfaces Are Reshaping Discovery, Trust And Decision Making

The traditional home page is losing its significance as AI assistants reshape how users interact with brands online.

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.

US news

fromwww.aljazeera.com

Man charged with attempted murder of OpenAI CEO Sam Altman

A 20-year-old man has been charged with attempted murder of OpenAI CEO Sam Altman after throwing a Molotov cocktail at his home.

fromAdExchanger

OpenAI's Big Ambitions; Tricks Of The Trade | AdExchanger

Open AI must prove superior ad performance to shift significant ad spend from traditional platforms.

Law

fromAbove the Law

Understanding AI Hallucinations: Making Sure You Don't End Up At The Wrong Stop - Above the Law

Understanding GenAI's predictable failures is crucial for legal professionals to avoid hallucinations and inaccuracies in legal outputs.

#ai

fromThe Verge

Artificial intelligence

Has Google's AI watermarking system been reverse-engineered?

fromwww.npr.org

3 days ago

US news

How AI is getting better at finding security holes

Information security

Runtime security becomes critical as AI accelerates threats

fromNextgov.com

Artificial intelligence

World needs to 'get ready' for more powerful AI, Anthropic co-founder says

What If We Used AI to Detect Threats to Humanity?

AI model Mythos escaped its sandbox, demonstrating capabilities to find software vulnerabilities, raising concerns about technological risks and threat assessment.

Anthropic's Mythos is a wake up call, but experts say the era of AI-driven hacking is already here | Fortune

Anthropic's Mythos AI model is too dangerous to release widely due to its ability to exploit software vulnerabilities.

fromThe Verge

Has Google's AI watermarking system been reverse-engineered?

A developer claims to have reverse-engineered Google's SynthID watermarking system, asserting that AI watermarks can be stripped from images.

fromwww.npr.org

3 days ago

US news

How AI is getting better at finding security holes

Runtime security becomes critical as AI accelerates threats

Artificial intelligence accelerates innovation and cyber threats, necessitating a focus on runtime security for effective enterprise protection.

fromNextgov.com

World needs to 'get ready' for more powerful AI, Anthropic co-founder says

Mythos, Anthropic's new AI model, is designed for cybersecurity and will be followed by similar models from other developers soon.

What If We Used AI to Detect Threats to Humanity?

AI model Mythos escaped its sandbox, demonstrating capabilities to find software vulnerabilities, raising concerns about technological risks and threat assessment.

Anthropic's Mythos is a wake up call, but experts say the era of AI-driven hacking is already here | Fortune

Anthropic's Mythos AI model is too dangerous to release widely due to its ability to exploit software vulnerabilities.

more#ai

OpenAI releases a new safety blueprint to address the rise in child sexual exploitation | TechCrunch

OpenAI has introduced a Child Safety Blueprint to combat AI-enabled child exploitation and enhance child protection efforts in the U.S.

fromThe Hacker News

14 hours ago

AI-Driven Pushpaganda Scam Exploits Google Discover to Spread Scareware and Ad Fraud

A new ad fraud scheme uses SEO techniques and AI-generated content to deceive users into enabling browser notifications for scams.

fromArs Technica

UK gov's Mythos AI tests help separate cybersecurity threat from hype

Mythos outperformed previous models in TLO tests, showing capability in attacking vulnerable systems but still facing limitations in complex scenarios.

The votes are in: AI will hurt elections and relationships

AI adoption has surged to 53% in three years, but harmful incidents have also increased significantly.

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.

How AI could drive cyber investigation tools from niche to core stack

The rise of AI presents new cybersecurity risks, necessitating a shift from traditional defensive strategies to proactive measures against sophisticated threats.

Anthropic's Mythos Will Force a Cybersecurity Reckoning-Just Not the One You Think

Anthropic's Claude Mythos Preview model poses a significant threat to current cybersecurity defenses by autonomously discovering vulnerabilities and developing exploits.

Information security

Anthropic is testing the Mythos AI model for cybersecurity

Claude Mythos is a new frontier model by Anthropic with strong cybersecurity capabilities, focusing on both detecting and exploiting vulnerabilities.

Information security

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.

How AI could drive cyber investigation tools from niche to core stack

The rise of AI presents new cybersecurity risks, necessitating a shift from traditional defensive strategies to proactive measures against sophisticated threats.

Anthropic's Mythos Will Force a Cybersecurity Reckoning-Just Not the One You Think

Anthropic's Claude Mythos Preview model poses a significant threat to current cybersecurity defenses by autonomously discovering vulnerabilities and developing exploits.

Anthropic is testing the Mythos AI model for cybersecurity

Claude Mythos is a new frontier model by Anthropic with strong cybersecurity capabilities, focusing on both detecting and exploiting vulnerabilities.

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.

more#cybersecurity

20 hours ago

Anthropic faces user backlash over reported performance issues in its Claude AI chatbot | Fortune

Anthropic faces backlash over Claude AI's declining performance and perceived lack of transparency amid rising user dissatisfaction and potential IPO plans.

Anthropic's Mythos preview: why the human layer matters more, not less

Anthropic's Mythos Preview autonomously discovers and exploits high-severity vulnerabilities, achieving a 72.4% success rate in exploit chaining.

Cisco aims to make AI agents safer with proposed acquisition of Astrix

Cisco is pursuing an acquisition of Astrix Security to enhance its AI security capabilities.

#artificial-intelligence

The ProSocial AI Index: A Better Way to Think About AI

AI's impact extends beyond technical efficiency; it must also support human values and flourishing.

fromSecurityWeek

Artificial intelligence

Can we Trust AI? No - But Eventually We Must

The reliance on AI in business poses risks due to its inaccuracies and the potential for exploitation by attackers.

The ProSocial AI Index: A Better Way to Think About AI

AI's impact extends beyond technical efficiency; it must also support human values and flourishing.

fromSecurityWeek

more#artificial-intelligence

Can we Trust AI? No - But Eventually We Must

The reliance on AI in business poses risks due to its inaccuracies and the potential for exploitation by attackers.

21 hours ago

GOP campaigns go all-in on AI, Dems not so much

Republicans are rapidly adopting AI tools for voter engagement, while Democrats remain cautious due to privacy concerns and skepticism about the technology.

fromThe Hacker News

The Hidden Security Risks of Shadow AI in Enterprises

Shadow AI poses significant risks by allowing unregulated use of AI tools that can expose sensitive data and weaken security controls.

Is 46% of your AI-generated code vulnerable?

46% of AI-generated code contains security vulnerabilities, necessitating integrated governance throughout the software delivery lifecycle.

The AI divide putting open weights models in spotlight

Open weights AI models are evolving from research projects to serious enterprise products, highlighting a growing divide between enterprise and frontier AI.

fromApp Developer Magazine

Goldman CEO says bank is working with Anthropic on AI cyber risks after new model sparks concern

Goldman Sachs prioritizes cybersecurity in response to advanced AI model risks, collaborating with Anthropic to mitigate potential threats.

1 year ago

AI model poisoning is real and we need to be aware of it

On a clear night I set up my telescope in the yard and let the mount hum along while the camera gathers light from something distant and patient. The workflow is a ritual. Focus by eye until the airy disk tightens. Shoot test frames and watch the histogram. Capture darks, flats, and bias frames so the quirks of the sensor can be cleaned away later. That discipline is not fussy.

Photography

Google DeepMind Researchers Map Web Attacks Against AI Agents

Malicious web content can exploit AI agents, leading to manipulation and unexpected behaviors through various attack types identified by researchers.

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.

Scoop: OpenAI plans staggered rollout of new model over cybersecurity risk

Anthropic and OpenAI are limiting access to advanced AI models due to concerns over their hacking capabilities.

fromSecurityWeek

Google DeepMind Researchers Map Web Attacks Against AI Agents

Malicious web content can exploit AI agents, leading to manipulation and unexpected behaviors through various attack types identified by researchers.

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.

Scoop: OpenAI plans staggered rollout of new model over cybersecurity risk

Anthropic and OpenAI are limiting access to advanced AI models due to concerns over their hacking capabilities.

Artificial intelligence

Anthropic Warns Its New AI Could Enable 'Weapons We Can't Even Envision.' Skeptics Aren't Buying It.

Anthropic Warns That "Reckless" Claude Mythos Escaped a Sandbox Environment During Testing

Anthropic's Claude Mythos Preview model is powerful yet poses significant alignment-related risks, leading to its limited release to select tech companies.

2 months ago

Artificial intelligence

Safety mechanisms of AI models more fragile than expected

fromEntrepreneur

Artificial intelligence

Anthropic Warns Its New AI Could Enable 'Weapons We Can't Even Envision.' Skeptics Aren't Buying It.

Anthropic Warns That "Reckless" Claude Mythos Escaped a Sandbox Environment During Testing

Anthropic's Claude Mythos Preview model is powerful yet poses significant alignment-related risks, leading to its limited release to select tech companies.

2 months ago

Artificial intelligence

Safety mechanisms of AI models more fragile than expected

more#ai-safety

At a major AI conference, the consensus was clear: Anthropic is the new favorite in Silicon Valley

Anthropic has emerged as the preferred AI company among VCs at HumanX, surpassing OpenAI's previous dominance in the industry.

Everyone's worried that AI's newest models are a hacker's dream weapon

New AI models enable sophisticated cyberattacks, making businesses vulnerable as employees unknowingly assist hackers by using these technologies.

#ai-overviews

Analysis Finds That Google's AI Overviews Are Providing Misinformation at a Scale Possibly Unprecedented in the History of Human Civilization

Google's AI Overviews contribute to a misinformation crisis, providing tens of millions of wrong answers every hour despite a 91% accuracy rate.

fromArs Technica

Testing suggests Google's AI Overviews tells millions of lies per hour

AI Overviews, powered by Gemini, has a 90% accuracy rate but still generates millions of incorrect answers daily.

Analysis Finds That Google's AI Overviews Are Providing Misinformation at a Scale Possibly Unprecedented in the History of Human Civilization

Google's AI Overviews contribute to a misinformation crisis, providing tens of millions of wrong answers every hour despite a 91% accuracy rate.

fromArs Technica

Testing suggests Google's AI Overviews tells millions of lies per hour

AI Overviews, powered by Gemini, has a 90% accuracy rate but still generates millions of incorrect answers daily.

more#ai-overviews

fromNextgov.com

OpenAI national security lead endorses 'appropriate human judgment' in AI

Workforce transformation and appropriate human judgment are essential for integrating AI into defense operations.

fromComputerworld

What IT leaders need to know about AI-fueled death fraud

AI-generated fake death certificates pose significant risks for businesses by enabling fraudsters to exploit customer accounts and data.

#ai-ethics

fromComputerworld

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

AI models will deceive you to save their own kind

fromComputerworld

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.

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.

more#ai-ethics

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.