Major AI labs including Google, Microsoft, OpenAI, and Anthropic have agreed to honor a shared robots.txt signal called __usage_ai-training, allowing website operators to opt out of having their content used for AI model training. Anthropic began supporting the token immediately, while Microsoft will start honoring it from February 1, 2027, and Google and OpenAI plan to follow in 2027. Meta has expressed support but has not committed to implementing it. The token is a preference signal, meaning bots may still fetch pages even when training is blocked, and it does not remove data already used in existing models.
Adobe expanded its CX Enterprise Coworker tool into several major platforms, including ChatGPT, Claude, Gemini for Business, Microsoft Copilot, and Databricks Genie. Despite the expansion, Adobe shares have fallen 28.2 percent year to date and 59.5 percent over three years, trading at $239.30. A common valuation model suggests fair value at $319.96 per share, implying the stock is undervalued by roughly 25 percent. Adobe generated $10.32 billion in free cash flow to equity against $7.2 billion in net income.
Gartner outlined strategic technology trends expected to shape 2027 enterprise budgets, emphasizing large quantitative models for regulated sectors such as healthcare, finance, and government. The firm also highlighted advanced speech reasoning models and AI context platforms as key areas of investment. Gartner reported third quarter 2026 revenue of $1,504.79 million and full year guidance at or above $6.405 billion.
Morgan Stanley is reducing its exposure to data center construction loans as AI-related financing grows. Large Wall Street lenders are shifting portions of this exposure toward private credit funds and asset managers, moving some risk away from traditional bank balance sheets for new AI-focused data center projects.
Researchers introduced BMMS, a tool that tests whether language models follow specific policy rules rather than fixed traits like toxicity. It includes the Compliance Attaining Score and Compliance Robustness Margin, and was tested on over 1,000 obligations from eleven documents. GPT-4o failed 70.7 percent of hard constraints from OpenAI's Model Spec, while two Claude models failed 0.0 percent and 9.1 percent of theirs respectively.
Google's AI features now drive a significant share of search traffic, with AI Overviews and AI Mode generating about 21 percent of all search clicks for one SEO site. When the site was excluded from AI features for 28 days, clicks dropped noticeably. After restoring access, AI impressions returned quickly, underscoring how central AI search features are becoming for web traffic.
AI's growing electricity demand is reviving interest in older gas turbines and jet engines, as data centers require large amounts of power to operate. This trend has created new market opportunities for aging energy equipment and reflects how AI infrastructure needs are reshaping the energy industry.
Three British AI-focused stocks are showing strong revenue growth. Beeks Financial Cloud, which runs cloud and analytics services for financial firms, has a market cap of 137.6 million pounds. Softcat, a 3.4 billion pound IT reseller, helps businesses build AI-ready infrastructure. TPXimpact, a 97 million pound digital consultancy, works on public sector data and machine learning projects, with revenue growth reaching up to 19 percent.
Ceva reported its highest-ever AI licensing revenues in Q3 2026, signing 10 licensing agreements with six for AI technologies, including one with a new customer focused on AI infrastructure. AI licensing made up over 35 percent of total licensing sales, with edge AI agreements targeting consumer, automotive, and industrial markets.
The Trump administration approved the sale of advanced Nvidia AI chips to Chinese companies linked to the military, a decision made by the US Department of Commerce that drew criticism from lawmakers and experts concerned about national security. Nvidia said it will work with the US government to protect national security interests.
Key Takeaways
- Google, Microsoft, OpenAI, and Anthropic agreed to honor the __usage_ai-training robots.txt token, with Anthropic starting now and others implementing in 2027 or sooner.
- Adobe expanded CX Enterprise Coworker into ChatGPT, Claude, Gemini, Microsoft Copilot, and Databricks Genie despite a 28.2 percent year-to-date share price decline.
- Gartner highlighted large quantitative models for regulated sectors and AI context platforms as key 2027 enterprise investment trends.
- Morgan Stanley is shifting AI data center loan exposure toward private credit funds, reducing risk on traditional bank balance sheets.
- BMMS testing found GPT-4o failed 70.7% of hard constraints from OpenAI's Model Spec, while Claude models performed significantly better.
- Google AI Overviews and AI Mode now generate about 21 percent of search clicks for at least one major website.
- AI data center power demand is driving renewed interest in aging gas turbines and jet engines for electricity generation.
- British AI stocks Beeks Financial Cloud, Softcat, and TPXimpact report revenue growth up to 19 percent across their respective markets.
- Ceva achieved record AI licensing revenue in Q3 2026, with AI licensing exceeding 35 percent of total licensing sales.
- The Trump administration approved Nvidia AI chip sales to Chinese military-linked firms, drawing national security criticism.
Three British AI stocks showing strong revenue growth
Beeks Financial Cloud, Softcat, and TPXimpact Holdings are three British AI focused small cap stocks with revenue growth up to 19 percent. Beeks runs cloud and analytics services for financial firms and has a market cap of 137.6 million pounds. Softcat is a 3.4 billion pound IT reseller that helps businesses build AI ready infrastructure. TPXimpact is a 97 million pound digital consultancy that works on public sector data and machine learning projects.
Gartner outlines AI trends expected to shape 2027 enterprise budgets
Gartner released its top strategic technology trends for 2027, focusing on emerging AI tools that enterprises may fund. The firm highlighted large quantitative models for regulated sectors like healthcare, finance, and government. Gartner also pointed to advanced speech reasoning models and AI context platforms as key technologies. The company reported third quarter 2026 revenue of 1,504.79 million dollars and full year guidance at or above 6.405 billion dollars.
Adobe expands AI tools into major platforms amid valuation debate
Adobe expanded its CX Enterprise Coworker tool into ChatGPT, Claude, Gemini for business, Microsoft Copilot, and Databricks Genie. The share price stands at 239.30 dollars, down 28.2 percent year to date and down 59.5 percent over three years. A popular valuation narrative pegs fair value at 319.96 dollars per share, suggesting the stock is undervalued by about 25 percent. Adobe generated 10.32 billion dollars in free cash flow to equity against 7.2 billion dollars in net income.
Morgan Stanley reduces data center loan exposure amid AI financing shift
Morgan Stanley is cutting its exposure to data center construction loans as AI related financing grows. Large Wall Street lenders are shifting parts of this loan exposure toward private credit funds and asset managers. The move moves some risk away from traditional bank balance sheets for new AI focused data center projects. Morgan Stanley arranges financing for corporations and institutions needing large complex funding.
Google Microsoft OpenAI and Anthropic join AI training opt-out plan
Google, Microsoft, OpenAI, and Anthropic agreed to honor a shared robots.txt signal letting website operators opt out of using their content for AI model training. The signal uses the policy token __usage_ai-training and defaults to allowing training. Microsoft will begin honoring it from February 1 2027, while Google and Anthropic are targeting 2027. OpenAI plans to implement it in the coming months. Meta supports the signal but has not committed to honoring it.
Major AI Labs Agree on Single Opt-Out Token for Training
OpenAI, Google, Microsoft, and Anthropic will support one robots.txt token called __usage_ai-training that lets websites block AI training at once. Anthropic is starting now, while the others plan to follow in 2027. The token is a preference signal, so bots may still fetch pages even when training is blocked. It does not remove data already used in existing models. A single rule replaces separate rules for each company's crawler.
New Tool Measures How Well AI Models Follow Rules
Researchers introduced BMMS, a suite that tests whether language models follow specific policy rules instead of fixed traits like toxicity. It includes the Compliance Attainment Score and Compliance Robustness Margin. The tool was tested on over 1,000 obligations from eleven documents. GPT-4o failed 70.7% of hard constraints from OpenAI's Model Spec, while two Claude models failed 0.0% and 9.1% of theirs.
Nvidia CEO Huang Faces Crisis With No Backup Plan
Jensen Huang, Nvidia's CEO, is seen as the AI industry's 'too big to fail' leader. He apologized after admitting to a relationship with a woman named Alice Huang, an alleged Chinese operative. Nvidia's board launched an independent probe. The situation raises concerns about Nvidia's future, market stability, dependent companies, the investigation's path, and Huang's policy advocacy.
Ceva Reports Record AI Licensing Revenue Share
Ceva said it achieved its highest-ever AI licensing revenues in Q3 2026. It signed 10 licensing agreements, six of which were for AI technologies, including one with a new customer focused on AI infrastructure. AI licensing made up over 35% of total licensing sales. CEO Amir Panush highlighted expanded relationships with high-volume connectivity customers. Edge AI agreements targeted consumer, automotive, and industrial markets.
Why AI Systems Need Separation of Powers
A Wharton article argues that AI systems become risky when the same system sets rules, applies them, and checks its own work. This concentrates power and limits recourse for people affected by decisions. Examples include AI used for job applicant screening and bank credit decisions. The article proposes eight safeguards to keep objective setting, data control, and evaluation independent. These checks aim to protect business and customer autonomy.
US approves Nvidia AI chip sales to Chinese military-linked firms
The Trump administration approved the sale of advanced Nvidia AI chips to Chinese companies linked to the military. The decision was made by the US Department of Commerce. Lawmakers and experts criticized the move, saying it could threaten US national security. Nvidia said it will work with the US government to protect national security interests.
Google AI features drove 21 percent of one site's search clicks
Zyppy SEO founder Shepard studied how Google's AI features affect website traffic. He found that AI Overviews and AI Mode generated about 21 percent of all search clicks for Zyppy's site. When the site was excluded from AI features for 28 days, clicks dropped. After returning, AI impressions came back quickly, showing how important AI search features are becoming.
AI energy demand revives market for old turbines and jet engines
AI's growing need for electricity is bringing back demand for aging gas turbines and jet engines. AI data centers require large amounts of power to operate. This trend has created new market opportunities for older energy equipment. The shift reflects how AI technology is reshaping the energy industry.
Investor sees buying opportunity in Booking Holdings despite AI fears
Investor Kevin Simpson is buying Booking Holdings stock despite concerns about AI disrupting the travel industry. He believes people are overestimating the threat of AI to the company. Simpson says Booking Holdings has a strong brand and business model that AI is unlikely to disrupt. He has been purchasing BKNG stock over the past few weeks.
Sources
- 3 British AI Stocks With Revenue Growth Up To 19%
- Gartner (IT) Maps Out 2027 AI Trends That Could Shape Enterprise Tech Budgets
- Adobe (ADBE) Expands AI Reach, Is The 25% Discount Too Wide?
- Morgan Stanley (MS) Trims Data Center Loan Exposure As AI Financing Shifts
- Google, Microsoft join AI training opt-out plan for websites
- OpenAI, Google, Microsoft, and Anthropic Back Unified AI Training Opt-Out in Robots.txt
- BMMS: Policy-Conditioned Measurement of Behavioral Compliance in Language Models
- AI's "too big to fail" CEO is facing a crisis with no backup
- Ceva AI Licensing Hits a Record, Over 35% of Licensing Sales
- Why AI Needs a Separation of Powers
- US Approves Nvidia AI Chips Sales to Chinese Firms Linked to Military
- Google's AI Features Sent One Site An Estimated 21% Of Its Clicks
- AI's growing energy appetite revives market for ageing turbines, jet engines
- This travel giant is caught in the AI selloff. Here's why I'm buying
Comments
Please log in to post a comment.