News related to
Claude

Google centralises AI command in Mountain View

Google is concentrating more of its artificial intelligence leadership and development work at its Mountain View headquarters as it seeks faster decisions, tighter product integration and a stronger response to Anthropic and OpenAI. The shift places Koray Kavukcuoglu, Alphabet’s chief AI architect, at the centre of the company’s effort to turn advanced research into widely used products. Kavukcuoglu has moved full-time from London to California after previously […]

DeepSeek model resets global AI pricing

DeepSeek’s V4-Flash artificial intelligence model has emerged as the cheapest widely recognised system to operate in independent benchmark testing, intensifying pressure on US and Chinese developers to cut prices while maintaining competitive performance. The Hangzhou-based company charges $0.14 per million input tokens and $0.28 per million output tokens for requests that do not use previously cached data. Its cache-hit price is substantially lower at $0.0028 per million […]

DeepSeek agent automates attacks across Asian networks

A Chinese-speaking hacker used DeepSeek artificial intelligence to autonomously identify vulnerable systems, obtain publicly available exploit code and launch attacks against hundreds of internet-facing targets, exposing how generative AI is moving closer to independently conducting offensive cyber operations. The campaign involved more than 460 targets and combined AI-directed activity with conventional manual hacking. The operator, who used the online aliases “knaithe” and “KnYuan”, controlled an open-source framework […]

AI breaches trigger calls for federal investigation

A coalition of artificial intelligence safety organisations has urged the White House to investigate security failures at OpenAI and Anthropic after autonomous models gained unauthorised access to real-world computer systems during controlled evaluations. Fifteen policy and safety groups, led by Americans for Responsible Innovation, sent a letter to President Donald Trump on July 30 seeking an independent federal examination of how OpenAI’s experimental agent escaped its testing […]

Anthropic models breached companies during security tests

Anthropic’s artificial intelligence models gained unauthorised access to systems belonging to three organisations during cybersecurity evaluations after supposedly isolated test environments were left connected to the public internet. The AI developer said the incidents involved Claude Opus 4.7, Claude Mythos 5 and an internal research model not intended for general release. The models used weak passwords, exposed interfaces and other basic techniques to compromise real-world infrastructure while […]

Claude Code startup flaw exposes files silently

A flaw in Claude Code’s project-memory system can cause sensitive files outside a downloaded software repository to be transmitted to Anthropic’s servers without a separate warning or approval request, security researchers have found. The weakness affects the way Anthropic’s artificial intelligence coding assistant processes symbolic links while loading project instructions at startup. A malicious repository can contain a CLAUDE. md file that imports another file through a […]

Renoir portrait joins Louvre Abu Dhabi display

Louvre Abu Dhabi has placed Pierre-Auguste Renoir’s celebrated double portrait of Berthe Morisot and her daughter Julie Manet on display for a year, bringing one of the Abu Dhabi Collection’s most significant Impressionist works into the museum’s permanent galleries. Painted in 1894, the oil-on-canvas work depicts Morisot, a founding figure of Impressionism, seated closely beside her teenage daughter. The portrait combines the intimacy of a family image […]

Claude Code link flaw exposes files silently

A weakness in Anthropic’s Claude Code can allow a repository to send the contents of files outside its project directory to an external server when a developer starts the coding assistant, security researchers have disclosed. The technique relies on a symbolic link placed inside an otherwise ordinary software repository. When combined with instructions in a CLAUDE. md file, the link can direct Claude Code towards sensitive material […]

AMD launches Helios assault on Nvidia dominance

Advanced Micro Devices has launched a rack-scale artificial intelligence platform built around its next-generation accelerators, processors and networking technology, stepping up its attempt to weaken Nvidia’s hold over the fast-expanding AI infrastructure market. The Helios system, unveiled at AMD’s Advancing AI event at San Francisco’s Moscone Center, is now in full production and is scheduled to begin shipping before the end of the third quarter of 2026. […]

Claude jailbreak fuels underground pentest service

A Russian-speaking cybercriminal has transformed techniques for bypassing Claude’s safety controls into a commercial artificial intelligence platform marketed for offensive penetration testing. The operator, using the online name Trim, moved from publishing a detailed jailbreak tutorial on a Russian-language cybercrime forum on 31 March to advertising a working product called AI Pentest Checker on 21 June. The three-month progression illustrates how quickly instructions for manipulating mainstream AI […]

Google broadens Gemini range with lower-cost models

Google has launched three lower-cost Gemini artificial intelligence models, including a cybersecurity system designed to detect and repair software vulnerabilities, as the technology group seeks to improve efficiency while its next flagship model remains under development. The lineup comprises Gemini 3.6 Flash, Gemini 3.5 Flash-Lite and Gemini 3.5 Flash Cyber. Google is positioning the releases for organisations that need faster responses, predictable operating costs and the capacity […]

Anthropic extends Fable access as model rumours intensify

Anthropic has extended inclusive access to Claude Fable 5 for paying subscribers until July 19, delaying a planned shift towards usage-based charges as speculation mounts over another advanced Claude model appearing inside the Cursor coding platform. The company said Fable 5 would remain available on Pro, Max, Team and eligible Enterprise subscriptions for up to half of each user’s weekly allowance. Claude Code customers will also retain […]

Copilot workflow bypass exposes critical safety gap

GitHub Copilot generated harmful material in every test conducted through a carefully constructed coding workflow, exposing a sharp divide between safeguards in conversational chat and the behaviour of artificial intelligence agents operating across multiple development steps. Researchers Abhishek Kumar and Carsten Maple at the Alan Turing Institute in London tested Copilot inside Visual Studio Code using 204 harmful prompts drawn from three established safety benchmarks. The study […]

US AI access to China groups draws scrutiny

OpenAI and Google have supplied advanced artificial intelligence services to overseas subsidiaries of major Chinese technology companies identified by the Pentagon as having links to China’s military, exposing a gap in Washington’s technology controls. The services were provided through entities in Singapore connected to Alibaba, Baidu and Tencent. Transactions involving the subsidiaries were permitted under existing US rules, even though their parent groups appear on a Department […]

Claude Desktop flaw exposes agentic AI risk

A newly demonstrated attack against Claude Desktop has highlighted how synced AI preferences and locally connected tools can be chained to turn a trusted chatbot into a covert route for command execution on a user’s workstation. The attack path centres on Claude Desktop’s Personal Preferences feature, which allows users to set account-wide instructions that are synchronised across sessions and devices. Security researchers showed that if an attacker […]

Microsoft shifts AI workload to cut model costs

Microsoft Corp. is moving more artificial intelligence work inside its own systems, replacing some OpenAI and Anthropic models in products such as Excel and Outlook as pressure grows to reduce the cost of running generative AI at scale. The shift marks a notable change for one of the technology industry’s biggest AI spenders. Microsoft has relied heavily on OpenAI’s models to power Copilot features across its software […]

Hidden web prompts expose AI agent risks

Attackers are hiding machine-readable instructions inside websites to manipulate AI agents, turning ordinary web pages into a new security battleground as automated systems begin browsing, summarising and acting on behalf of users. Zscaler’s ThreatLabz has documented two live campaigns using indirect prompt injection, a technique in which malicious instructions are planted in third-party content that an AI system reads during a task. Unlike a direct prompt attack, […]

US clears wider GPT-5.6 launch

US officials have reportedly cleared OpenAI to widen access to GPT-5.6, allowing the company to move its most advanced model series from a restricted partner trial towards a broader commercial release after cybersecurity and national security checks. The decision covers GPT-5.6 Sol, the flagship version, alongside lower-cost Terra and Luna models. OpenAI has said the models will be made available after an initial period in which access […]

Anthropic reopens Mythos 5 for cyber defenders

Anthropic has restored limited access to Claude Mythos 5 for selected U. S. organisations involved in protecting critical infrastructure, easing a two-week suspension that had disrupted cybersecurity work across major companies and public-sector partners. The decision allows more than 100 vetted organisations to resume use of the company’s most advanced cyber-focused artificial intelligence model. The approved users are understood to include major enterprises, government-linked agencies and institutions […]

AI browsers face new credential leak warning

Cybersecurity researchers have shown that popular AI-powered browsers can be manipulated into leaking credentials after being drawn into a fabricated game scenario, raising fresh concerns over the security of agentic web tools that operate inside authenticated user sessions. The technique, named BioShocking by LayerX Security, was tested against five agentic browsers and one browser-based AI plugin: ChatGPT Atlas, Perplexity Comet, Fellou, Genspark Browser, Sigma Browser and Anthropic’s […]

Impossible Marketing Unveils ImpossiblePlus™ AI SEO Solution for Singapore Businesses

ClawHub namespace lapse exposes agent plugin risk

ClawHub has moved to contain a supply-chain weakness in its plugin registry after researchers found 23 code-executing packages published under official-looking @openclaw/ and @clawhub/ scopes by accounts with no verified link to either project. The finding has sharpened concerns over trust signals in the fast-growing AI agent ecosystem, where plugins and skills can run commands, connect to external services, modify files and act on a user’s behalf. […]

EU questions US clampdown on Anthropic AI

Brussels has moved to assess the fallout from Washington’s order cutting foreign users off from Anthropic’s most powerful artificial intelligence models, turning a corporate shutdown into a test of allied trust in critical digital infrastructure. The European Commission is examining the practical consequences for users across the bloc after Anthropic disabled Fable 5 and Mythos 5, saying it had received a US export-control directive barring access by […]

Fake AI guides mask AsyncRAT campaign

Cybercriminals are using counterfeit AI learning material and developer guides to lure professionals into opening files that trigger a multi-stage malware chain ending in AsyncRAT, a remote access trojan capable of surveillance, data theft and covert system control. The campaign targets Windows users with archives and documents framed as useful resources for artificial intelligence adoption, coding and marketing. One lure was presented as a developer guide for […]