The FBI says cybercriminals and state-linked hackers are using ARTEX, an AI-powered penetration-testing framework, to automate attacks across finance, government, technology, health care and energy. In some cases, it can help attackers find targets, exploit vulnerabilities, gain privileged access and steal data in under three hours. The FBI says the tool has been used since at least August 2026, including in attacks attributed to China-linked groups. Reporting also links ARTEX to the October ConnectWise Cloud Backup breach, which exposed Microsoft 365 customers to attacks that Microsoft connected to Storm-0940. Amazon reported ARTEX use in an attempt to steal AI credentials, while Indiaâs CERT-In identified the group Crafty Falcon as targeting banks, fintech firms and cryptocurrency exchanges. Experts warn that chatbot-style prompts and a simple interface can make sophisticated attacks accessible to less-skilled users. The FBI urges organizations to strengthen defenses and monitor for suspicious automated activity.
Google is adding temporary chats to AI Mode in Search. These sessions do not become part of a userâs usual search history, are detached from the account, and do not use past preferences or shape future personalization. The feature is rolling out worldwide on desktop and mobile in supported languages. Google Images is also adding a Lens button that can direct image-search follow-ups to AI Overviews or AI Mode. Separately, a free Kaggle and Google course teaches learners to build agents with Gemini and Googleâs ADK, covering tools such as Google Search and MCP, agent workflows, memory, and evaluation using logs, traces and metrics.
Anthropic is reportedly seeking up to $90.4 billion in an IPO, exceeding the $75 billion SpaceX raised in June. Founded in 2021 by seven former OpenAI employees, the company makes the Claude models and is led by CEO Dario Amodei and President Daniela Amodei. About 20% of 825 million shares are expected to be offered to retail investors in several countries. A Bloomberg report said the offering was roughly six times oversubscribed, and Harvest ETFs and Ninepoint Partners plan Anthropic-focused funds. In a separate disclosure, Anthropic reported $900.9 billion in aggregate compute commitments, with $70.7 billion due over the next 12 months. Analyst Dan Ives said the commitments offer chipmakers and cloud providers a public demand signal, while questions remain about industry returns and Anthropicâs future margins.
Dell says rising demand for AI models and distributed inference is pushing computing beyond conventional data centers. It is shipping what it calls the first fully liquid-cooled networking system, with switches, optics and cables tested and delivered together in a rack; Dell says this can reduce installation from weeks to hours. Its software plans to connect as many as 1,000 smaller sites, drawing on spare electricity or methane to run AI workloads. Morgan Stanleyâs Gokul Laroia, meanwhile, sees potential investment value in Chinaâs independent AI ecosystem, biotech advances and industrial AI adoption. He cautions that tighter U.S. chip restrictions could weigh in the near term, though Chinaâs semiconductor self-sufficiency drive may lessen the effect over time.
Other reports focus on putting AI to work and managing its costs. One article says strict employee usage limitsâincluding a reported Tesla cap of $200 per weekâcan make developers postpone using AI tools when they could help with current work. It recommends matching models to tasks and routing suitable routine coding to less expensive options. A report on Databricksâ Genie One says it reached 95% accuracy with governed context; definition questions reached 100%, and document-plus-data questions reached 97% when agents used that context. In a retail example, governed context lifted accuracy from 54% to 82%.
McDonaldâs is expanding tests of its ArchIQ AI drive-thru platform alongside a Home Alone-themed meal, a planned nationwide value menu and a specialty-beverage push. The article frames AI ordering and beverage promotions as tools the company may use to improve labor efficiency and emphasize higher-margin sales. McDonaldâs reported third-quarter revenue of US$7,157 million and net income of US$2,486 million, and issued two US$600 million bond offerings due in 2031 and 2036.
Key Takeaways
- The FBI says ARTEX has enabled automated attacks since at least August 2026, with some reaching privileged access in under three hours; Microsoft linked attacks affecting Microsoft 365 customers after the ConnectWise breach to China-aligned Storm-0940.
- Amazon reported ARTEX use in an attempt to steal AI credentials; Indiaâs CERT-In identified Crafty Falcon as targeting banks, fintech firms and crypto exchanges.
- Google is rolling out account-detached temporary chats in AI Mode worldwide and adding a Google Lens pathway from Images to AI Overviews or AI Mode.
- A free Kaggle-Google course teaches AI-agent development with Gemini and ADK, including tools, workflows, memory and evaluation.
- Anthropic is reportedly seeking up to $90.4 billion in an IPO; a Bloomberg report said the offering was roughly six times oversubscribed.
- Anthropic disclosed $900.9 billion in aggregate compute commitments, including $70.7 billion due in the next 12 months.
- Dell says its fully liquid-cooled networking rack can cut installation from weeks to hours and plans software to connect up to 1,000 smaller computing sites.
- Morgan Stanley sees investment potential in Chinaâs AI ecosystem but warns U.S. chip restrictions could have near-term effects.
- An article cites a reported Tesla employee AI-use cap of $200 per week and argues for matching model costs to tasks; Databricksâ Genie One reportedly reached 95% accuracy with governed context.
- McDonaldâs is expanding ArchIQ drive-thru tests while promoting new offers; Q3 revenue was US$7,157 million and net income was US$2,486 million.
FBI warns AI hacking tool can reach privileged access in under three hours
The FBI warned on November 13, 2026, that cybercriminals and state-linked groups are using ARTEX, an AI-powered penetration-testing framework, to automate attacks. The tool can find targets, exploit weaknesses, gain privileged access and steal data, sometimes in less than three hours. The FBI says attackers have used ARTEX since at least August 2026 against organizations in sectors including finance, government, technology, health care and energy. One reported incident involved the October 2026 ConnectWise Cloud Backup breach, which exposed Microsoft 365 customers to attacks linked by Microsoft to China-aligned group Storm-0940. The FBI urges organizations to strengthen security and watch for suspicious activity.
ARTEX makes sophisticated AI-assisted hacking easier for beginners
ARTEX is an AI-powered penetration-testing tool that lets users scan systems for weaknesses and guide attacks through chatbot prompts. The article says its simple interface has made it popular with cybercriminals who may lack advanced technical skills. Cybersecurity experts and businesses are concerned that the tool could make sophisticated attacks easier to launch. ARTEX is one of several AI-powered tools described, alongside DeepTest, which uses machine learning to find possible vulnerabilities.
FBI warns China-linked hackers are using ARTEX in attacks
The FBI says hackers linked to China have allegedly used the AI-powered penetration-testing framework ARTEX in attacks on banks and other organizations. The article also describes an attack on ConnectWise and an Amazon report that hackers used ARTEX while trying to steal AI credentials. Indiaâs CERT-In identified the group as Crafty Falcon, which it said targeted banks, financial technology firms and cryptocurrency exchanges. The FBI says ARTEX can help users with little technical skill run complex operations against many targets, with some attacks reaching privileged access in under three hours. It urged organizations to improve security and watch for suspicious automated activity.
Google adds temporary chats to AI Mode in Search
Google is adding temporary chats to AI Mode in Search, giving users a way to have a conversation that does not become part of their usual search history. Users can start one by clicking the temporary threads icon in the upper right corner of AI Mode. Google says the session is detached from the account, so its responses will not use past preferences or affect future personalization. The feature is rolling out worldwide on desktop and mobile in all supported languages.
Google Lens in Images can lead to AI Overviews and AI Mode
Google Images results now include a Google Lens button that can open a follow-up query in AI Overviews or AI Mode. Damien Andell shared a video showing the feature in action. The article says that trying the button outside the Google app opened Google AI Overviews. The change shows Google directing more image-search activity into its AI features.
What Kaggleâs Free Google Course Teaches About Building AI Agents
A free course from Kaggle and Google teaches people to build AI agents with Gemini and Googleâs ADK. It covers agent design, tools such as Google Search and MCP, and ways to connect agents in sequences, groups, or loops. The lessons also explain how sessions and long-term memory help agents keep useful context, and how to evaluate their work with logs, traces, and metrics. Hands-on notebooks guide learners through the topics and discuss running agents beyond a notebook.
Morgan Stanley Sees Investment Potential in Chinaâs AI Ecosystem
Morgan Stanleyâs Gokul Laroia said Chinaâs independent AI ecosystem, biotech advances, and use of AI in industry could attract new investment and create long-term value. He warned that tighter U.S. chip restrictions could have a near-term effect. However, he expects Chinaâs drive for semiconductor self-sufficiency to reduce the impact over time.
Dell Looks to Turn Spare Power Into Distributed AI Computing
Dell says growing AI models and demand for distributed inference are pushing AI factories beyond traditional data centers. The company is shipping what it calls the first fully liquid-cooled networking system, with switches, optics, and cables tested and delivered together in a rack. Dell says this setup can cut installation from weeks to hours. Its software plans aim to connect as many as 1,000 smaller sites, using spare power sources such as store electricity or methane to run AI computing.
Anthropicâs Planned IPO Could Raise a Record $90.4 Billion
Anthropic is seeking to raise up to US$90.4 billion in an IPO, which would top the record US$75 billion raised by SpaceX in June. The company was founded in 2021 by seven former OpenAI employees and is led by CEO Dario Amodei and President Daniela Amodei. It makes the Claude family of AI models, including Haiku, Sonnet, Opus, and Fable, and operates as a public benefit corporation focused on responsible AI development. The company expects about 20 percent of 825 million shares to go to retail investors in several countries, including Canada, the United States, Australia, Japan, and the United Kingdom. A Bloomberg report on Nov. 12 said the offering was roughly six times oversubscribed, while Canadian firms Harvest ETFs and Ninepoint Partners plan Anthropic-focused ETFs.
Why Strict AI Spending Limits Can Hurt Employees
Some companies limit employee access to AI tools to control costs, including a reported Tesla limit of $200 per week. The article says these caps can force developers to save tool use for later rather than use it when it could help with current work. It argues that companies should match AI tools to the task instead of relying on strict usage limits. Routine coding work may not need the most expensive model, and tools can automatically send tasks to lower-cost models that can handle them well.
Anthropicâs AI Commitments Offer a Major Demand Signal
Anthropic disclosed $900.9 billion in aggregate compute commitments in a November 10 amendment, with $70.7 billion due in the next 12 months. Tech analyst Dan Ives of Yorkville Ives said the commitments give chipmakers and cloud providers a public signal of demand. Questions remain about whether the AI industry can earn enough to support its high spending. Investors will also be watching Anthropicâs future profit margins as it grows.
Genie One Reaches 95% Accuracy With Governed Context
A post reported that Databricksâ Genie One reached 95% accuracy using governed context. Before that, four team agents for Finance, Reserving, Underwriting and Claims each scored above 85% on basic questions, while tables alone scored 64%. Genie One had to work across tables, 14 PDFs and conflicting information, including planted traps. Definition questions reached 100% accuracy with the governed layer, and document plus data questions reached 97% with agents and that layer. In retail, adding governed context raised accuracy from 54% to 82%.
McDonaldâs Expands ArchIQ AI Drive Thru Tests Alongside New Offers
McDonaldâs is expanding AI drive thru testing through its ArchIQ platform as it rolls out holiday and long-term initiatives. These include a Home Alone themed meal, a planned nationwide value menu and a push for specialty beverages. The article says AI ordering and beverage promotions may help the company improve labor efficiency and focus on higher margin sales as costs and demand change. McDonaldâs reported Q3 revenue of US$7,157 million and net income of US$2,486 million. The company also issued two US$600 million bond offerings, due in 2031 and 2036, which could support technology and other plans while keeping its debt level in focus.
Sources
- FBI Warns Businesses Worldwide of AI Tool Enabling Cyberattacks in Under Three Hours
- Criminals Have a New Favorite âPoint-and-Shootâ AI Hacking Tool
- China-linked hackers using AI model ARTEX to infiltrate banks, FBI warns
- Google's AI feature in Search gets its own incognito mode
- Google Images Results With Lens To AI Mode Feature
- What I Learned From Kaggle + Google's Free Agentic AI Course
- China's independent AI ecosystem to draw capital despite U.S. chip curbs: Morgan Stanley
- Distributed inference turns stranded power into AI compute
- Can Anthropicâs IPO live up to the hype?
- 4 places where cutting AI costs is a mistake
- Anthropic Pushes the Reset Button for AI
- Genie One hits 95% with governed context
- How Investors Are Reacting To McDonald's (MCD) AI Drive Thru Expansion
Comments
Please log in to post a comment.