#ai-security

[ follow ]
#data-protection
more#data-protection
#cybersecurity
Information security
fromITPro
3 months ago

Public sector workers are sweating over AI security threats

A significant number of public sector IT professionals express concerns over AI's security implications, particularly regarding data privacy and compliance.
Information security
fromZero Day Initiative
3 months ago

Zero Day Initiative - Announcing Pwn2Own Berlin and Introducing an AI Category

Pwn2Own 2025 will be hosted at OffensiveCon in Berlin, introducing a new AI category focused on advanced security challenges.
The inclusion of AI as a category reflects the growing concerns and interest in the security of AI technologies.
Artificial intelligence
fromInfoQ
3 months ago

Google Gemini's Long-term Memory Vulnerable to a Kind of Phishing Attack

Johann Rehberger demonstrated a prompt injection attack on Google Gemini, exploiting delayed tool invocation to modify its long-term memories.
Information security
fromITPro
3 months ago

Public sector workers are sweating over AI security threats

A significant number of public sector IT professionals express concerns over AI's security implications, particularly regarding data privacy and compliance.
Information security
fromZero Day Initiative
3 months ago

Zero Day Initiative - Announcing Pwn2Own Berlin and Introducing an AI Category

Pwn2Own 2025 will be hosted at OffensiveCon in Berlin, introducing a new AI category focused on advanced security challenges.
The inclusion of AI as a category reflects the growing concerns and interest in the security of AI technologies.
Artificial intelligence
fromInfoQ
3 months ago

Google Gemini's Long-term Memory Vulnerable to a Kind of Phishing Attack

Johann Rehberger demonstrated a prompt injection attack on Google Gemini, exploiting delayed tool invocation to modify its long-term memories.
more#cybersecurity
Artificial intelligence
fromZDNET
2 weeks ago

Microsoft unveils new AI agent customization and oversight features at Build 2025

Microsoft is advancing personalized AI through new features like Copilot Tuning and Multi-Agent Orchestration.
Enterprises can customize AI tools without coding, reflecting their unique operations.
#cybersecurity-risks
more#cybersecurity-risks
#prompt-injection
Artificial intelligence
fromFuturism
1 month ago

Researchers Find Easy Way to Jailbreak Every Major AI, From ChatGPT to Claude

A newly discovered jailbreak can manipulate AI models into producing harmful content, exposing vulnerabilities in their safety measures.
Artificial intelligence
fromInfoQ
1 month ago

DeepMind Researchers Propose Defense Against LLM Prompt Injection

Google DeepMind's CaMeL effectively neutralizes 67% of prompt injection attacks in LLMs using traditional software security principles.
fromInfoQ
3 weeks ago
Artificial intelligence

Meta Open Sources LlamaFirewall for AI Agent Combined Protection

Artificial intelligence
fromFuturism
1 month ago

Researchers Find Easy Way to Jailbreak Every Major AI, From ChatGPT to Claude

A newly discovered jailbreak can manipulate AI models into producing harmful content, exposing vulnerabilities in their safety measures.
Artificial intelligence
fromInfoQ
1 month ago

DeepMind Researchers Propose Defense Against LLM Prompt Injection

Google DeepMind's CaMeL effectively neutralizes 67% of prompt injection attacks in LLMs using traditional software security principles.
fromInfoQ
3 weeks ago
Artificial intelligence

Meta Open Sources LlamaFirewall for AI Agent Combined Protection

more#prompt-injection
Marketing tech
fromInfoQ
3 weeks ago

Meta Announces API and Protection Tools at First LlamaCon Event

Meta's LlamaCon introduced new AI development tools including the Llama API and protection features aimed at empowering developers.
#artificial-intelligence
Artificial intelligence
fromZDNET
3 weeks ago

Multimodal AI poses new safety risks, creates CSEM and weapons info

Multimodal AI enhances LLMs but increases their vulnerability to novel attacks.
New research indicates significant safety risks with multimodal models, exposing them to dangerous outputs.
Artificial intelligence
fromZDNET
3 weeks ago

Multimodal AI poses new safety risks, creates CSEM and weapons info

Multimodal AI enhances LLMs but increases their vulnerability to novel attacks.
New research indicates significant safety risks with multimodal models, exposing them to dangerous outputs.
more#artificial-intelligence
#zero-trust
Privacy professionals
fromITPro
2 months ago

Microsoft launches new security AI agents to help overworked cyber professionals

Microsoft is enhancing its Security Copilot with new AI agents to support IT teams facing rising security threats.
Privacy professionals
fromITPro
2 months ago

Microsoft launches new security AI agents to help overworked cyber professionals

Microsoft is enhancing its Security Copilot with new AI agents to support IT teams facing rising security threats.
more#zero-trust
Artificial intelligence
fromIT Pro
1 month ago

"There needs to be an order of magnitude more effort": AI security experts call for focused evaluation of frontier models and agentic systems

AI risk evaluation requires significant improvement as systems evolve rapidly.
Experts call for enhanced security measures to keep pace with evolving AI technologies.
NYC politics
fromPCMAG
1 month ago

NYC Bets on AI Surveillance to Clean Up Subways, Predict Criminal Behavior

MTA is leveraging AI to enhance subway security by identifying potential threats in real-time using video surveillance.
Artificial intelligence
fromIT Pro
1 month ago

"Governance is an irreplaceable role": Microsoft Security VP on why diversity and sector expertise will keep security workers relevant in the age of agentic AI

AI skills and diversity are vital for security teams in a rapidly evolving threat landscape.
#software-development
European startups
fromTechCrunch
1 month ago

Endor Labs, which builds tools to scan AI-generated code for vulnerabilities, lands $93M | TechCrunch

AI-generated code brings new security challenges, with many organizations facing issues frequently.
Endor Labs shifted focus to address vulnerabilities in AI-created code, leading to significant funding success.
fromZDNET
1 month ago
Artificial intelligence

1Password extends enterprise credential management beyond humans to AI agents

European startups
fromTechCrunch
1 month ago

Endor Labs, which builds tools to scan AI-generated code for vulnerabilities, lands $93M | TechCrunch

AI-generated code brings new security challenges, with many organizations facing issues frequently.
Endor Labs shifted focus to address vulnerabilities in AI-created code, leading to significant funding success.
fromZDNET
1 month ago
Artificial intelligence

1Password extends enterprise credential management beyond humans to AI agents

more#software-development
Artificial intelligence
fromtime.com
1 month ago

Exclusive: Every AI Datacenter Is Vulnerable to Chinese Espionage, Report Says

U.S. AI datacenters are at significant risk of espionage and sabotage, threatening national security and tech investments.
#funding
Startup companies
fromSilicon Canals
2 months ago

Botguard rebrands to Blackwall, lands 45M to scale AI-powered security: Know more - Silicon Canals

Blackwall raises €45M to expand AI-based security solutions for SMBs and plans significant growth in the US and APAC markets.
Startup companies
fromSilicon Canals
2 months ago

Botguard rebrands to Blackwall, lands 45M to scale AI-powered security: Know more - Silicon Canals

Blackwall raises €45M to expand AI-based security solutions for SMBs and plans significant growth in the US and APAC markets.
more#funding
Privacy professionals
fromHackernoon
2 months ago

The HackerNoon Newsletter: Is Your AI-Generated Code Really Secure? (3/20/2025) | HackerNoon

AI tools are transforming developers' workflows, enhancing productivity and security concerns.
Political and legal challenges in Argentina are intensified by a significant crypto scandal.
#espionage
Privacy professionals
fromTechCrunch
2 months ago

Anthropic CEO says spies are after $100M AI secrets in a 'few lines of code' | TechCrunch

Dario Amodei warns about algorithmic espionage by China targeting U.S. AI firms, urging government intervention.
Privacy professionals
fromTechCrunch
2 months ago

Anthropic CEO says spies are after $100M AI secrets in a 'few lines of code' | TechCrunch

Dario Amodei warns about algorithmic espionage by China targeting U.S. AI firms, urging government intervention.
more#espionage
Information security
fromITPro
2 months ago

12,000 API keys and passwords were found in a popular AI training dataset - experts say the issue is down to poor identity management

The exposure of nearly 12,000 valid secrets in AI training datasets reveals significant vulnerabilities in identity management practices.
Artificial intelligence
fromBusiness Insider
4 months ago

US Navy email warns against using AI apps like China's DeepSeek

The US Navy has warned against using the Chinese AI app DeepSeek due to security concerns.
A memo sent to Navy personnel is a reminder of existing policy against open-source AI tools.
Artificial intelligence
fromInfoWorld
5 months ago

The vital role of red teaming in safeguarding AI systems and data

Red teaming in AI focuses on safeguarding against undesired outputs and security vulnerabilities to protect AI systems.
Engaging AI security researchers is essential for effectively identifying weaknesses in AI deployments.
[ Load more ]