#generative-ai-security
#generative-ai-security

fromThe Hill

Can AI agents protect our privacy?

AI is transforming the web, increasing privacy risks while offering opportunities to improve data protection.

Science

The Pluripotent Ocean of Emerging AI

fromFast Company

Artificial intelligence

No, McDonald's AI bot didn't go rogue, but 'prompt injection' is still a risk for companies

The Horrible Economics of AI Are Starting to Come Crashing Down

Tech leaders are now rating employee productivity based on AI token usage, signaling a shift in AI access and costs.

AI Can Autonomously Hack Cloud Systems With Minimal Oversight: Researchers

AI systems can autonomously hack cloud environments, demonstrating advanced capabilities in executing sophisticated attacks without specific instructions.

2 hours ago

Experts Warn of AI Swarms Hijacking Democracy With Fake Citizens

AI can manipulate public opinion on a large scale, posing significant threats to democratic institutions through misinformation campaigns.

fromThe Hill

Can AI agents protect our privacy?

AI is transforming the web, increasing privacy risks while offering opportunities to improve data protection.

Science

The Pluripotent Ocean of Emerging AI

Human attachments to language model chatbots mirror the uncanny experiences of scientists with the ocean on Solaris, leading to psychological consequences.

fromFast Company

No, McDonald's AI bot didn't go rogue, but 'prompt injection' is still a risk for companies

Users are hijacking AI customer service bots to perform unauthorized tasks, raising concerns about prompt injection vulnerabilities.

The Horrible Economics of AI Are Starting to Come Crashing Down

Tech leaders are now rating employee productivity based on AI token usage, signaling a shift in AI access and costs.

AI Can Autonomously Hack Cloud Systems With Minimal Oversight: Researchers

AI systems can autonomously hack cloud environments, demonstrating advanced capabilities in executing sophisticated attacks without specific instructions.

Hot take: AI's not going to kill open source code security

Cal.com has shifted from AGPL-3.0 to a proprietary license, raising concerns about open source security in the AI era.

How to Know Where Your Security Threat Is Before It's Too Late

Organizations winning the security talent war operationalize key questions to prevent knowledge loss and enhance cybersecurity resilience.

Discord Sleuths Gained Unauthorized Access to Anthropic's Mythos

Mozilla used Anthropic's Mythos Preview to fix 271 vulnerabilities in Firefox 150, while North Korean hackers exploited AI for cybercrime.

Mythos access by Discord group reveals real danger of AI-powered hacking | Fortune

Unauthorized access to AI models exposes vulnerabilities in cybersecurity, highlighting the rapid pace of AI-driven exploitation of flaws.

Information security

A group of users leaked Anthropic's AI model Mythos by reportedly guessing where it was located | Fortune

fromThe Cipher Brief

Anthropic Mythos - We've Opened Pandora's Box

Cybersecurity faces a new threat from AI systems like Anthropic Mythos, which can exploit vulnerabilities and shift the balance in favor of attackers.

Information security

Chinese Cybersecurity Firm's AI Hacking Claims Draw Comparisons to Claude Mythos

Careers

fromEntrepreneur

How to Know Where Your Security Threat Is Before It's Too Late

Organizations winning the security talent war operationalize key questions to prevent knowledge loss and enhance cybersecurity resilience.

Discord Sleuths Gained Unauthorized Access to Anthropic's Mythos

Mozilla used Anthropic's Mythos Preview to fix 271 vulnerabilities in Firefox 150, while North Korean hackers exploited AI for cybercrime.

Mythos access by Discord group reveals real danger of AI-powered hacking | Fortune

Unauthorized access to AI models exposes vulnerabilities in cybersecurity, highlighting the rapid pace of AI-driven exploitation of flaws.

A group of users leaked Anthropic's AI model Mythos by reportedly guessing where it was located | Fortune

Unauthorized access to Anthropic's Mythos model raises significant cybersecurity concerns.

fromThe Cipher Brief

Anthropic Mythos - We've Opened Pandora's Box

Cybersecurity faces a new threat from AI systems like Anthropic Mythos, which can exploit vulnerabilities and shift the balance in favor of attackers.

Intellectual property law

Chinese Cybersecurity Firm's AI Hacking Claims Draw Comparisons to Claude Mythos

A Chinese cybersecurity firm claims AI-driven vulnerability discovery capabilities rivaling Anthropic's Claude Mythos model.

more#cybersecurity

#generative-ai

4 hours ago

Devious New AI Tool "Clones" Software So That the Original Creator Doesn't Hold a Copyright Over the New Version

Generative AI challenges copyright by using copyrighted material without permission, creating tools that bypass existing licenses.

Marketing tech

fromAP News

Intellectual property law

AI is a gold mine for spammers and scammers, but Google is using it as a tool to fight back

Generative AI tools have intensified online spam and scams, prompting tech companies like Google to enhance their defenses against malicious ads.

4 hours ago

Devious New AI Tool "Clones" Software So That the Original Creator Doesn't Hold a Copyright Over the New Version

Generative AI challenges copyright by using copyrighted material without permission, creating tools that bypass existing licenses.

Marketing tech

fromAP News

AI is a gold mine for spammers and scammers, but Google is using it as a tool to fight back

Generative AI tools have intensified online spam and scams, prompting tech companies like Google to enhance their defenses against malicious ads.

Information security

Why Cybersecurity Must Rethink Defense in the Age of Autonomous Agents

Software development

fromDevOps.com

Agentic AI for Defense: How Checkmarx Turns Security into a Coding Partner - DevOps.com

Agentic AI proactively identifies and addresses security vulnerabilities in real-time during code development, enhancing application security significantly.

Government adoption of AI agents could outpace the private sector

Agentic AI adoption in government is a leadership mandate, with 82% already using it and 71% planning to increase usage by 2026-2027.

Why Cybersecurity Must Rethink Defense in the Age of Autonomous Agents

Agentic AI is transforming cybersecurity, presenting both opportunities for defenders and risks for attackers, necessitating a strategic response from the industry.

Software development

fromDevOps.com

Agentic AI for Defense: How Checkmarx Turns Security into a Coding Partner - DevOps.com

Agentic AI proactively identifies and addresses security vulnerabilities in real-time during code development, enhancing application security significantly.

Government adoption of AI agents could outpace the private sector

Agentic AI adoption in government is a leadership mandate, with 82% already using it and 71% planning to increase usage by 2026-2027.

Mowing Down Simulated Elephants Could Help Self-Driving Cars Prepare For the Chaos of Real Life Streets

New benchmark for testing self-driving cars introduces random scenarios to improve model robustness and address limitations in current training methods.

OpenAI's Sam Altman apologizes for not reporting ChatGPT account of Tumbler Ridge suspect to police

Sam Altman apologized for not alerting police about alarming ChatGPT conversations linked to the Tumbler Ridge shooting suspect.

Florida AG opens criminal investigation into OpenAI and ChatGPT

Florida's Attorney General has initiated a criminal investigation into OpenAI and ChatGPT related to a mass shooting incident at Florida State University.

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.

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.

Canada news

18 hours ago

OpenAI's Sam Altman apologizes for not reporting ChatGPT account of Tumbler Ridge suspect to police

Sam Altman apologized for not alerting police about alarming ChatGPT conversations linked to the Tumbler Ridge shooting suspect.

Florida AG opens criminal investigation into OpenAI and ChatGPT

Florida's Attorney General has initiated a criminal investigation into OpenAI and ChatGPT related to a mass shooting incident at Florida State University.

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.

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.

Prestigious Wall Street Law Firm Humiliated When Its AI Use Is Discovered in Court

AI hallucinations in legal filings can lead to significant professional embarrassment and potential sanctions.

Startup companies

Your Former Employer Is Selling Your Slacks and Emails to Train AI

Founders of defunct startups are monetizing their digital remains, such as Slack messages and emails, through a growing ecosystem of buyers and middlemen.

fromTNW | Opinion

Business intelligence

How web intelligence is powering the next wave of AI Infrastructure

The web intelligence industry is evolving to support AI's growing demands for multimodal data processing, particularly in handling video content.

DevOps

fromTechRepublic

2 years ago

What is Cloud Security? Fundamental Guide

Cloud security requires specialized processes and technologies to protect assets and data from evolving threats in a dynamic environment.

Online marketing

fromIndependent

Why your AI assistant is suddenly selling to you

Sponsored chats are transforming digital advertising by integrating promotions into conversations with large language models.

Women in technology

The military embraced AI long before Claude

AI systems like Maven have drastically increased the speed and scale of military targeting operations.

fromTNW | Health-Tech

Healthcare

How AI Is Reshaping Workers' Compensation Claims and Healthcare Operations

Workers' compensation is a significant yet often overlooked part of the healthcare ecosystem, facing unique challenges and requiring focused innovation.

Data science

fromInfoWorld

5 days ago

Addressing the challenges of unstructured data governance for AI

Enterprises must enhance data governance for unstructured data as AI transforms data management practices.

Digital life

fromSilicon Canals

5 days ago

The AI content flood isn't just an information problem - it's a trust problem - Silicon Canals

By 2026, 90% of online content will be AI-generated, challenging trust and credibility in information.

6 hours ago

How to audit what ChatGPT knows about you - and reclaim your data privacy

Re-evaluate personal information shared with ChatGPT to enhance privacy and security.

#ai-security

fromwww.aljazeera.com

Artificial intelligence

Who's in control of AI?

What Claude and OpenClaw Vulnerabilities Reveal About AI Agents

AI agents must be governed like privileged accounts due to their ability to access sensitive information and execute commands.

Information security

How indirect prompt injection attacks on AI work - and 6 ways to shut them down

Information security

Anthropic's Mythos breach was humiliating

fromwww.theguardian.com

The Guardian view on Anthropic's Claude Mythos: when AI finds every flaw, who controls the internet? | Editorial

Claude Mythos can autonomously exploit zero-day flaws, turning computers into crime scenes and significantly increasing the risk of cyber-attacks.

Anthropic's most dangerous AI model just fell into the wrong hands

Mythos AI model accessed by unauthorized users, raising cybersecurity concerns about its potential misuse.

fromwww.aljazeera.com

Who's in control of AI?

Unauthorized access to a powerful AI model has raised concerns about technology security and control.

What Claude and OpenClaw Vulnerabilities Reveal About AI Agents

AI agents must be governed like privileged accounts due to their ability to access sensitive information and execute commands.

How indirect prompt injection attacks on AI work - and 6 ways to shut them down

Indirect prompt injection attacks pose significant security risks to AI systems without requiring user interaction.

Anthropic's Mythos breach was humiliating

Unauthorized access to Anthropic's AI model Mythos raises serious concerns about cybersecurity and safety protocols.

fromwww.theguardian.com

The Guardian view on Anthropic's Claude Mythos: when AI finds every flaw, who controls the internet? | Editorial

Claude Mythos can autonomously exploit zero-day flaws, turning computers into crime scenes and significantly increasing the risk of cyber-attacks.

Anthropic's most dangerous AI model just fell into the wrong hands

Mythos AI model accessed by unauthorized users, raising cybersecurity concerns about its potential misuse.

more#ai-security

Business intelligence

fromEntrepreneur

The Hidden Data Liability Every Leader Needs to Address Now

Data is no longer endlessly renewable; companies face a 'data liability gap' affecting AI systems and data recovery responsibilities.

fromThe Hacker News

Bridging the AI Agent Authority Gap: Continuous Observability as the Decision Engine

AI agents create a delegation gap in enterprise security that must be addressed before governing their authority.

Why are top university websites serving porn? It comes down to shoddy housekeeping.

Universities often neglect DNS record maintenance, leading to hijacked subdomains that can appear in search results.

Software development

fromInfoWorld

Microsoft taps Anthropic's Mythos to strengthen secure software development

Mythos can enhance the security of Microsoft products, benefiting enterprises without direct access.

Business intelligence

Scaling agentic AI demands a strong data foundation - 4 steps to take first

Trusted quality data is essential for scaling agentic AI adoption in organizations.

fromComputerworld

CISA last in line for access to Anthropic Mythos

CISA lacks access to Anthropic's AI model Claude Mythos, while unauthorized users have gained access, raising cybersecurity concerns.

#artificial-intelligence

How Should Effective AI Red Teams Operate?

AI-specific red teaming is essential for organizations to understand and mitigate risks associated with AI tools.

fromInfoQ

Deepfakes, Disinformation, and AI Content Are Taking Over the Internet

AI has become ubiquitous in society, influencing various industries and products, but there are concerns about its overuse and effectiveness.

AI and the Unintended Self

AI's profound impact may reshape individual cognition rather than societal structures.

5 AI Models Tried to Scam Me. Some of Them Were Scary Good

Artificial intelligence is increasingly capable of executing sophisticated social engineering attacks, as demonstrated by the DeepSeek-V3 model.

Former national cyber director: Anthropic's 'Mythos' AI can hack nearly anything and we aren't ready | Fortune

Mythos, Anthropic's advanced AI model, poses significant risks to critical infrastructure, necessitating urgent investment and collaboration to enhance cybersecurity.

The Mythos meeting focused on the wrong AI risk to banks. Here's the one nobody is talking about | Fortune

Artificial intelligence is transforming fraud into a machine-driven, scalable threat, posing risks beyond traditional cyber attacks.

How Should Effective AI Red Teams Operate?

AI-specific red teaming is essential for organizations to understand and mitigate risks associated with AI tools.

fromInfoQ

Deepfakes, Disinformation, and AI Content Are Taking Over the Internet

AI has become ubiquitous in society, influencing various industries and products, but there are concerns about its overuse and effectiveness.

AI and the Unintended Self

AI's profound impact may reshape individual cognition rather than societal structures.

5 AI Models Tried to Scam Me. Some of Them Were Scary Good

Artificial intelligence is increasingly capable of executing sophisticated social engineering attacks, as demonstrated by the DeepSeek-V3 model.

Former national cyber director: Anthropic's 'Mythos' AI can hack nearly anything and we aren't ready | Fortune

Mythos, Anthropic's advanced AI model, poses significant risks to critical infrastructure, necessitating urgent investment and collaboration to enhance cybersecurity.

more#artificial-intelligence

The Mythos meeting focused on the wrong AI risk to banks. Here's the one nobody is talking about | Fortune

Artificial intelligence is transforming fraud into a machine-driven, scalable threat, posing risks beyond traditional cyber attacks.

fromFast Company

Meta tracking employee keystrokes to train AI is probably legal. Experts say that doesn't make it ethical

Meta Platforms is implementing software to track employee computer usage to train AI models, raising privacy concerns amid potential layoffs.

The Privacy-Security Partnership: How We Bend Risk in a Resource Crunch

Fewer privacy practitioners feel confident in meeting laws, while resource shortages and compliance challenges increase stress in the field.

fromTNW | Business

Florida opens first criminal AI probe into OpenAI

Florida Attorney General opens investigation into OpenAI for ChatGPT's alleged role in a mass shooting.

fromIT Brew

Asking around: When does ransomware threat intelligence become noise?

Effective threat intelligence requires filtering information relevant to specific market segments to avoid overwhelming alerts.

#meta

Meta Installing Software on Employee Computers to Track Everything They Do, Feed the Data to AI

Meta is implementing tracking software on employees' computers to gather data for AI training, raising ethical and privacy concerns.

Meta will use employee-tracking software to help train AI agents: Report

Meta will track US employees' mouse movements and clicks to improve AI training data.

Meta Installing Software on Employee Computers to Track Everything They Do, Feed the Data to AI

Meta is implementing tracking software on employees' computers to gather data for AI training, raising ethical and privacy concerns.

Meta will use employee-tracking software to help train AI agents: Report

Meta will track US employees' mouse movements and clicks to improve AI training data.

more#meta

#ai-in-cybersecurity

Wiz founder: Hack yourself with AI, before the bad guys do | Computer Weekly

Security leaders should proactively use AI tools on their systems to gain an advantage against cyber threats.

fromTheregister

Google unleashes even more AI security agents to fight crims

AI is essential for cybersecurity, transitioning from human-led to AI-led defense strategies overseen by humans.

Wiz founder: Hack yourself with AI, before the bad guys do | Computer Weekly

Security leaders should proactively use AI tools on their systems to gain an advantage against cyber threats.

fromTheregister

Google unleashes even more AI security agents to fight crims

AI is essential for cybersecurity, transitioning from human-led to AI-led defense strategies overseen by humans.

more#ai-in-cybersecurity

fromwww.businessinsider.com

The AI race is quietly rewriting what surveillance looks like at work

Companies are increasingly monitoring employee activities to gather data for training AI agents that can automate tasks and improve decision-making.

fromTechzine Global

Agentic AI is reshaping the network - and it's time to upgrade

Wireless connectivity is essential for AI, transforming industries and requiring strategic management to address complexity and security risks.

Now, even ransomware is using post-quantum cryptography

Kyber's use of PQC key-exchange algorithms serves more as a marketing tactic than a practical security measure against imminent quantum threats.

#anthropic

Anthropic's growing pains mount ahead of OpenAI showdown

Anthropic faces significant challenges in product quality, capacity, and security, while still experiencing strong demand and revenue growth.

Anthropic is investigating 'unauthorized access' of its Mythos cybersecurity tool

Anthropic is investigating unauthorized access to its Claude Mythos model, which is designed to identify cybersecurity flaws.

Anthropic's growing pains mount ahead of OpenAI showdown

Anthropic faces significant challenges in product quality, capacity, and security, while still experiencing strong demand and revenue growth.

Anthropic is investigating 'unauthorized access' of its Mythos cybersecurity tool

Anthropic is investigating unauthorized access to its Claude Mythos model, which is designed to identify cybersecurity flaws.

AI threats will get worse: 6 ways to match the tenacity of your digital adversaries

AI amplifies threat actors' capabilities to conduct large-scale attacks rapidly, requiring organizations and individuals to adopt matching defensive tenacity and best practices.

Google Antigravity in Crosshairs of Security Researchers, Cybercriminals

Google Antigravity's vulnerabilities have attracted both security researchers and cybercriminals, leading to risks of remote code execution and malware delivery.

#ai-training

Now Meta will track what employees do on their computers to train its AI agents

Meta is using employee activity data to train AI agents for better task automation.

fromTechCrunch

Meta will record employees' keystrokes and use it to train its AI models | TechCrunch

Meta is using employee data, including mouse movements and keystrokes, to train its AI models for improved efficiency.

Now Meta will track what employees do on their computers to train its AI agents

Meta is using employee activity data to train AI agents for better task automation.

fromTechCrunch

Meta will record employees' keystrokes and use it to train its AI models | TechCrunch

Meta is using employee data, including mouse movements and keystrokes, to train its AI models for improved efficiency.

more#ai-training

Anthropic's Mythos raises the stakes for security validation | Computer Weekly

The rise of autonomous AI in security introduces unpredictability, complicating the validation of defenses against evolving threats.

fromNextgov.com

Microsoft to test third-party AI models for incorporation in its security offerings

Microsoft is evaluating third-party AI systems to enhance its cybersecurity measures against AI-driven threats.

A tsunami of flaws: When frontier AI and Patch Tuesday collide | Computer Weekly

April 2025 Patch Tuesday update was the second-largest in history, addressing over 160 vulnerabilities, with AI tools potentially driving the increase.

6 days ago

58% of Organizations Spend Over 10 Hours a Month Securing AI-generated Code

31% of organizations using AI-generated code spend 10 hours or less per month on validation and auditing, raising security concerns.

fromSitePoint Forums | Web Development & Design Community

Why Your OpenClaw Setup is a "Malicious Insider" in Waiting

OpenClaw's capabilities pose significant security risks, revealing vulnerabilities that can be exploited by malicious insiders.

6 days ago

Anthropic's Mythos AI model sparks fears of turbocharged hacking

AI-enabled cyber attacks surged 89% in 2025, with attackers acting within 29 minutes of gaining access.

2 months ago

Why We Can't Let AI Take the Wheel of Cyber Defense

Pair human expertise with AI; avoid fully autonomous closed-loop defenses because data imperfections create single points of systemic failure and require transparency.