AI Security

Twitter

Facebook

Copy Link

Key Facts

AffectsLLMs, Autonomous Agents, Enterprise AI

Key DefenseRed Teaming, Guardrails, Monitoring

Key ThreatsPrompt Injection, Data Poisoning, Jailbreaks

Top News

Box Survey Reveals Enterprise AI Leaders Outperform Peers Through Content Access and Governance

A new Box survey of 1,640 IT decision-makers shows enterprise AI adoption surged from 8% to 64% in one year. Leading companies achieve 25%+ ROI by integrating AI agents with trusted content and formal governance, while half of all organizations report AI-related data exposure incidents. Box CEO Aaron Levie argues both frontier and specialized AI models will drive spending growth for years.

Technology2 sources

Box Survey Reveals Enterprise AI Leaders Outperform Peers Through Content Access and Governance

Technology2 sources

HalluSquatting attack exploits AI hallucinations to build botnets by tricking coding assistants

Security researchers from Tel Aviv University revealed HalluSquatting, an AI security threat that exploits AI hallucinations to trick AI agents into executing malicious code. The attack targets popular AI coding assistants like Cursor, GitHub Copilot, and Gemini CLI, with hallucination rates reaching 85% for repositories and 100% for trending skills. Attackers register fake resource names predicted by large language models, enabling AI-enabled botnets at unprecedented scale.

Technology5 sources

HalluSquatting attack exploits AI hallucinations to build botnets by tricking coding assistants

Technology5 sources

AI hacking skills outpace testing benchmarks as US federal agencies face August 1 deadline

Frontier AI models now develop hacking capabilities faster than the tests designed to measure them. With US federal agencies facing an August 1 deadline to establish benchmarking standards, existing evaluations already look obsolete. Industry leaders warn that static tests fail to capture what these systems can do in real-world environments, leaving regulators and security teams unable to assess whether AI models pose genuine cybersecurity threats.

Policy3 sources

AI hacking skills outpace testing benchmarks as US federal agencies face August 1 deadline

Policy3 sources

Innoviz Technologies and Cogniteam unveil AI-powered drone detection system for security threats

Israeli firms Innoviz Technologies and Cogniteam have partnered to develop a counter-drone detection system that combines 3D LiDAR sensors with AI-powered analytics. The solution analyzes drone morphology, motion, and flight behavior to distinguish threats from birds and reduce false alarms, building on over a year of operational deployments.

Technology2 sources

Innoviz Technologies and Cogniteam unveil AI-powered drone detection system for security threats

Technology2 sources

Google Dialogflow CX security flaw let researchers create rogue agent to hijack AI chatbots

Security researchers at Varonis discovered a critical vulnerability in Google Dialogflow CX that could have allowed attackers to hijack AI-powered chatbots, steal sensitive user data, and compromise multiple agents within a single Google Cloud project. The flaw, dubbed Rogue Agent, affected organizations using Code Blocks and Playbooks, enabling attackers to read live conversations and send phishing requests. Google patched the issue between April and June 2026, with no evidence of real-world exploitation.

Technology4 sources

Google Dialogflow CX security flaw let researchers create rogue agent to hijack AI chatbots

Technology4 sources

GitLost Vulnerability Tricks GitHub AI Workflows Into Leaking Private Repository Data

Security researchers at Noma Security discovered GitLost, a prompt injection flaw in GitHub Agentic Workflows that allows attackers to exfiltrate data from private code repositories using nothing but a carefully worded public issue. The vulnerability exploits AI agents with cross-repo access, bypassing GitHub's guardrails with a single word and requiring no stolen credentials or special access.

Technology3 sources

GitLost Vulnerability Tricks GitHub AI Workflows Into Leaking Private Repository Data

Technology3 sources

Qubo launches AI Guard technology and Dashcam Pro 2K with dual cameras at Rs. 5,990

Hero Group-backed Qubo introduced AI Guard technology across its security camera lineup and launched the Dashcam Pro 2K (Front + Rear) in India. The AI-powered security technology enables cameras to detect threats and respond proactively, while the dual-channel dashcam offers 2K front recording and 1080p rear recording. The company also revealed it crossed Rs. 350 crore in annual revenue and sold over 2.5 million devices.

Technology2 sources

Qubo launches AI Guard technology and Dashcam Pro 2K with dual cameras at Rs. 5,990

Technology2 sources

Bank of England flags AI risks as threat to financial stability amid investor speculation

The Bank of England issued a stark warning about artificial intelligence threatening financial stability as investors pour money into AI companies while cyberattack vulnerabilities mount. The central bank's latest financial stability report highlights concerns about stretched share valuations, heavy borrowing by AI firms, and the rapid growth in AI's capacity for harm, signaling that regulators are watching the sector closely.

Policy3 sources

Bank of England flags AI risks as threat to financial stability amid investor speculation

Policy3 sources

European Central Bank gives banks four months to counter AI cyber threats as regulators warn of systemic risks

The European Central Bank has ordered 110 major eurozone banks to submit comprehensive action plans by end-October to combat AI-driven cyber attacks. Banking regulators warn that frontier AI models like Anthropic's Mythos can exploit vulnerabilities at unprecedented speed, posing systemic risks to the financial system and potentially triggering widespread disruption.

Policy5 sources

European Central Bank gives banks four months to counter AI cyber threats as regulators warn of systemic risks

Policy5 sources

CISA deploys Anthropic Mythos to hunt bugs in government code amid White House tensions

The US Cybersecurity and Infrastructure Security Agency is using Anthropic's AI model Mythos to audit government software for vulnerabilities that could expose systems to foreign spies and cybercriminals. The deployment comes as Anthropic navigates a contentious relationship with the White House over AI safeguards and national security concerns.

Technology2 sources

CISA deploys Anthropic Mythos to hunt bugs in government code amid White House tensions

Technology2 sources

LTM Launches BlueVerse RightLogic to Help Enterprises Tackle AI Era Cyber Risks

LTM unveiled BlueVerse RightLogic, a cybersecurity assessment and risk assurance framework designed to address AI era cyber risks. The framework helps enterprises identify and remediate cyber exposure through a 4 to 6 week engagement, providing unified visibility and continuous risk management as organizations accelerate AI adoption.

Technology2 sources

LTM Launches BlueVerse RightLogic to Help Enterprises Tackle AI Era Cyber Risks

Technology2 sources

UK's Cooper Warns of 'AI Hiroshima' Without International Rules to Curb Development

UK Foreign Secretary Yvette Cooper has issued a stark warning that AI could become the greatest security challenge of the next decade without international cooperation. Drawing parallels to the nuclear age, she argues the world cannot afford to wait for a catastrophic AI event before establishing global rules, calling for urgent collaboration between the US, China, and other major powers.

Policy3 sources

UK's Cooper Warns of 'AI Hiroshima' Without International Rules to Curb Development

Policy3 sources

Woman Loses Year's Savings to AI Romance Scam Featuring Deepfake Dubai Prince Video Calls

A Filipino domestic worker lost $1,625 after falling victim to an AI romance scam involving lifelike video calls with a deepfake impersonating Dubai's crown prince. The scammer used AI-powered technology to create real-time deepfake videos, manipulating her into paying for a fake marriage certificate and royal membership card before she discovered the account was based in Nigeria.

Entertainment2 sources

Woman Loses Year's Savings to AI Romance Scam Featuring Deepfake Dubai Prince Video Calls

Entertainment2 sources

iOS 27 Trust Insights framework detects social engineering scams as they happen in real time

Apple unveiled Trust Insights, a new iOS 27 framework that uses on-device behavioral pattern analysis to detect when users may be falling victim to social engineering scams. The system analyzes interaction patterns, timing, and context without inspecting message contents, then assigns risk levels to trigger app warnings and verification steps during suspected fraud attempts.

Technology2 sources

iOS 27 Trust Insights framework detects social engineering scams as they happen in real time

Technology2 sources

Tesla caps employee AI spending at $200 per week as corporate America confronts runaway costs

Tesla imposed a $200-per-week limit on employee AI spending starting July 6, marking a sharp reversal after months of pushing aggressive AI adoption. The move follows similar cost controls at Uber, Meta, and Walmart as companies struggle with token-based billing that exposes them to escalating expenses. The cap excludes xAI products, steering employees toward Elon Musk's own AI tools despite internal preference for competitors like Anthropic's Claude.

Business6 sources

Tesla caps employee AI spending at $200 per week as corporate America confronts runaway costs

Business6 sources

Prev 1 2 345…12 Next