European Central Bank Vice President Boris Vujcic says elevated AI valuations and sovereign-debt concerns have become more acute risks to euro-zone financial stability. A sharp reversal in investor sentiment could drive steep price falls and losses, with regional investment funds’ growing AI exposure adding to the danger. He also points to rising debt-funded investment, concentrated exposure, rapid adoption and potential cyberattacks. US IPO activity is on track to exceed dot-com-boom levels, a pattern Vujcic says has often coincided with market peaks.
Vujcic urges banks to remain resilient and supervisors to monitor non-bank firms and private credit, where limited transparency, closer ties to banks and insurers, and possible liquidity mismatches raise concerns. Separately, a report on the $9.5 trillion foreign-exchange market says AI and round-the-clock trading are testing whether the voluntary FX Global Code can keep pace. CIBC’s Manuel Mondedeu says the Code may need more frequent reviews; a 30-member technology group is studying potential updates. New York Fed official Anna Nordstrom says AI could improve trading but reduce transparency, while continuous trading and blockchain-based instant settlement could mean thinner liquidity and greater volatility. The Global Foreign Exchange Committee is also seeking wider adoption of the Code.
At Commvault’s SHIFT event in Nashville on November 10, 2026, the company announced enterprise AI collaborations with NVIDIA and Databricks, plus an AI-security partnership with Zscaler. The NVIDIA work combines Commvault Data Activate with NVIDIA NIM microservices and NemotronRetriever; a reference design is planned for early release in December 2026, with general availability in 2027. A Databricks solution will share governed data tables through OpenSharing, with early access expected in select Azure regions in December. The Zscaler partnership targets AI visibility, policy enforcement and recovery, with no availability date announced.
Onyx Security and Glean are integrating agent discovery with runtime security controls. Onyx can review prompts, responses and proposed tool actions against company policies before they reach business systems, helping distinguish permission to retrieve information from permission to use or share it. Security teams can track agents, access, capabilities and risk-relevant changes, supported by audit records. The need for oversight is underscored by Palo Alto Networks’ 2026 State of Cloud Security Report, based on 2,800 security workers and decision-makers in 10 countries: 99% of organizations use AI agents for partly or fully independent software-development tasks, and 73% let agents change code, production settings and security tests. Organizations use fewer than three of eight measured pre-release checks on average; 19% of severe and critical vulnerabilities reach production, two in five organizations cannot reconstruct an agent’s actions, and 98% worry advanced models could find and exploit vulnerabilities independently.
StudioX AI says its Agent2Agent support allows agents on different platforms to exchange tasks and results without internet access. Teams can share an assistant or multi-agent Mission as an A2A agent, described by an Agent Card. Air-gapped deployments require the necessary models, identity tools and data sources inside the isolated network; teams can also build tools to review results, track actions and require human approval for selected decisions. Meanwhile, Hudson Rock reports that a criminal marketplace is using AI-written profiles to help buyers search stolen computer data in plain language. Listings may include passwords, cookies, autofill data and workplace VPN or Citrix credentials. The operator claims tens of millions of logs and over a billion parsed accounts, with prices from $0.25, but Hudson Rock has not verified those claims or confirmed completed purchases. An advertised Canadian eye-clinic workstation listing was priced at $10.
Corporate AI spending is rising despite reservations about letting systems act independently. In a survey of IT professionals, 63% of companies increased AI spending and 91% said projects met or exceeded their original goals; all respondents considered agentic AI important. Yet 77% reported internal hesitation about AI acting without human approval, while 90% of IT leaders want autonomous IT operations within two years. Riverbed CIO Fernando Castanheira recommends starting with low-risk uses and maintaining strong security practices. Guidance on AI portfolio-review tools likewise stresses human checks: the tools can identify concentration, ETF overlap and information in SEC filings, but may misread data or miss important context, and should not replace human judgment. Investors are advised to verify findings with experts and share documents through secure platforms with access limited to authorized people.
China’s AI hardware downturn is putting growth-focused mutual funds under pressure. Changsheng Growth Select entered liquidation after its net asset value remained below 50 million yuan—about $7.5 million—for 50 working days; it was down 33.27% for the year as of November 13. On November 15, BOC Growth Select Equity A, Boshi Quality Select Mixed and Penghua Zhitou Digital Economy Mixed A issued warnings after heavy losses. Their AI-hardware holdings and potential investor withdrawals could further shrink fund sizes, with continued volatility and no clear recovery putting more funds at risk. In London, Opportunity London CEO Jace Tyrrell says AI is drawing billions of dollars in investment and boosting demand for data centres, offices, laboratories, energy infrastructure and innovation districts. He also discusses automation’s effects on banking and investment opportunities as Australia and the UK pursue closer economic ties in finance, technology and infrastructure.
Key Takeaways
- ECB Vice President Boris Vujcic warns that high AI valuations, sovereign-debt concerns and a possible investor reversal pose growing euro-zone stability risks.
- Vujcic cites debt-funded AI investment, concentrated exposure, cyber risks and private-credit transparency and liquidity concerns; he calls for resilient banks and close supervision.
- Commvault’s NVIDIA collaboration combines Data Activate with NVIDIA NIM microservices and NemotronRetriever; a reference design is planned for December 2026 and general availability in 2027.
- Commvault and Databricks plan governed data sharing through OpenSharing, with early access in select Azure regions expected in December 2026; its Zscaler partnership targets AI visibility, policy enforcement and recovery.
- Onyx Security and Glean are integrating policy checks for agent prompts, responses and proposed actions, with tracking and audit records for enterprise oversight.
- Palo Alto Networks’ survey found 99% of organizations use AI agents in software development and 73% allow changes to code, production settings and security tests; two in five cannot reconstruct an agent’s actions.
- StudioX AI’s A2A support is designed to let agents exchange tasks without internet access, including in air-gapped setups that host required models and services internally.
- Hudson Rock found a criminal marketplace using AI-written profiles to organize stolen computer data; its operator’s claims of tens of millions of logs and over a billion accounts remain unverified.
- China’s Changsheng Growth Select entered liquidation below 50 million yuan, about $7.5 million, after 50 working days; three other AI-hardware-exposed funds issued warnings after losses.
- Opportunity London CEO Jace Tyrrell says AI is drawing billions in investment and raising demand for London data centres, offices, laboratories and energy infrastructure; separate surveys and reports highlight both growing AI investment and concerns about oversight, FX-market rules and reliance on AI portfolio tools.
ECB flags AI valuations and sovereign debt as key stability risks
European Central Bank Vice President Boris Vujcic said artificial intelligence and sovereign debt worries are prominent risks to euro zone financial stability. The ECB has long warned about debt sustainability in weaker euro zone countries and high AI stock valuations. Vujcic said these concerns have become more acute over the past year and need policymakers’ attention.
ECB warns an AI market reversal could trigger major losses
European Central Bank Vice President Boris Vujcic warned that a sudden change in investor views on AI could cause sharp market price drops and major losses. He said rising AI investment and high valuations have increased financial stability risks in the euro zone. Vujcic also noted that US IPO activity is on track to exceed levels seen during the dot-com boom, a pattern he said has often matched market peaks. Increased exposure to AI among investment funds in the region could make a crash a major risk for the euro zone.
ECB warns rapid AI growth adds risks to financial stability
European Central Bank Vice President Vujcic said rapid AI growth could threaten financial stability, as high AI stock valuations raise the risk of a market correction. He pointed to growing use of debt funding, concentrated exposure, faster AI adoption and possible cyberattacks as added risks. Vujcic said banks need to remain resilient and supervisors should keep watch over non-bank firms. He also called for close monitoring of private credit because of limited transparency, stronger links with banks and insurers, and possible mismatches in liquidity.
Commvault teams with NVIDIA, Databricks and Zscaler on enterprise AI
Commvault announced collaborations with NVIDIA and Databricks to prepare protected enterprise data for AI use, and partnered with Zscaler on AI security. The announcements were made at Commvault’s SHIFT event in Nashville on November 10, 2026. The NVIDIA work will combine Commvault Data Activate with NVIDIA NIM microservices and NemotronRetriever, while a reference design is planned for early release in December 2026 and general availability in 2027. A planned Databricks solution will share governed data tables through OpenSharing, with early access expected in select Azure regions in December 2026. The Zscaler work focuses on AI visibility, policy enforcement and recovery, but no availability date was given.
Onyx Security and Glean join forces to govern enterprise AI agents
Onyx Security and Glean are integrating agent discovery and runtime security controls for enterprise AI. Glean’s hooks let Onyx review prompts, responses and proposed tool actions against company policies before actions reach business systems. The controls are designed to distinguish permission to retrieve information from permission to use or share it. Security teams can track agents, their access and capabilities, and changes that may affect their risk, while audit records help teams review activity and policy enforcement.
AI agents spread across development, often without clear oversight
Palo Alto Networks’ 2026 State of Cloud Security Report surveyed 2,800 security workers and decision-makers in 10 countries. It found that 99 percent of organizations use AI agents for partly or fully independent tasks in software development, and 73 percent allow agents to change code, production settings, and security tests. While 82 percent said AI has led to more cloud updates, organizations use fewer than three of eight measured checks before release on average. The report also found that 19 percent of severe and critical vulnerabilities reach production, while two in five organizations cannot reconstruct what an agent did. The survey said 98 percent worry about advanced AI models that could find and exploit vulnerabilities on their own.
China AI hardware fund losses push several products toward liquidation
China’s AI hardware downturn has put several growth-focused mutual funds near liquidation. Changsheng Growth Select entered liquidation after its net asset value stayed below 50 million yuan, about $7.5 million, for 50 working days; it was down 33.27 percent this year as of Nov. 13. On Nov. 15, BOC Growth Select Equity A, Boshi Quality Select Mixed, and Penghua Zhitou Digital Economy Mixed A also issued warnings after heavy losses. Their portfolios included AI hardware companies, and the warnings may lead investors to withdraw money, putting further pressure on fund sizes. The report says continued market volatility without a clear recovery could put more funds at risk.
London draws billions as AI fuels demand for infrastructure
London is attracting billions of dollars in AI investment as the technology changes industries and workplaces. On The Connector, hosted by Belinda Coates, Opportunity London CEO Jace Tyrrell discussed the city’s appeal to global investors. He said AI is driving demand for data centres, offices, laboratories, energy infrastructure, and innovation districts, while automation is changing banking. The program also explored investment opportunities as Australia and the UK seek closer economic ties in finance, technology, and infrastructure.
StudioX AI connects agents across isolated networks
StudioX AI says its Agent2Agent, or A2A, support lets AI agents on different platforms exchange tasks and results without internet access. A StudioX assistant or multi-agent workflow called a Mission can be shared as an A2A agent, with an Agent Card explaining the tasks it can handle. Compatible agents can then send it work and receive results. In an air-gapped setup, the agents and services they need, including models, identity tools, and data sources, must be available inside the isolated network. StudioX says teams can also build an application with tools to review results, track agent actions, and require human approval for selected decisions.
Criminal marketplace uses AI profiles to sort stolen computer data
Hudson Rock found a criminal marketplace advertising stolen computer data alongside AI-written profiles of the machines’ owners and likely valuable accounts. Buyers can search the profiles in plain language, and the marketplace also offers filters such as location, operating system, and stealer malware type. The stolen logs may include passwords, cookies, autofill data, and workplace VPN or Citrix logins. The operator claims to hold tens of millions of logs and more than a billion parsed accounts, with prices starting at $0.25, but Hudson Rock has not verified those figures or confirmed completed purchases. One advertised listing offered data from a Canadian eye clinic workstation for $10, while other examples included computers linked to a defense contractor’s household and a government employee.
AI Investment Grows Despite Low Trust Among Companies
A survey of IT professionals found that 63% of companies increased AI spending, even as hesitation about the technology rose. Ninety-one percent said AI projects met or exceeded their original goals, and all respondents considered agentic AI important. Still, 77% reported internal hesitation about letting AI act without human approval, while 90% of IT leaders want to move toward autonomous IT operations within two years. Riverbed CIO Fernando Castanheira advised businesses to start with low-risk uses and maintain strong security practices.
AI and Round-the-Clock Trading Put Currency Market Rules to the Test
The rapid growth of AI and 24-hour trading is raising questions about whether rules can keep pace with changes in the $9.5 trillion foreign-exchange market. Manuel Mondedeu of CIBC Capital Markets said the voluntary FX Global Code may need more frequent reviews, while keeping its core principles. A 30-member technology working group is studying whether some rules need updates, though some market participants say the current standards already cover new technology. Federal Reserve Bank of New York official Anna Nordstrom said AI could improve trading but may reduce transparency, while round-the-clock trading and instant settlement using blockchain and tokenization could bring thinner liquidity and greater volatility. The Global Foreign Exchange Committee is also working to expand the Code’s use among smaller firms and groups such as trading venues, corporate treasurers and hedge funds.
What AI Can and Cannot Tell You About Your Investment Portfolio
AI-powered portfolio review tools can check investment concentration, overlap among ETFs and SEC filings. They can also offer insights into an investment strategy and suggest possible improvements. However, AI may miss the emotional side of investing, misread data or overlook important information, so it cannot replace human judgment. Investors should verify AI-generated insights with human experts before making decisions. The article also recommends using a secure online platform to share investment documents so only authorized people can access them.
Sources
- ECB VP says AI, debt have become prominent stability risks
- ECB warns of market risks if AI boom stalls
- ECB Warns Rapid AI Expansion Threatens EU Financial Stability
- Commvault links protected enterprise data to NVIDIA and Databricks and teams with Zscaler on AI security
- Onyx Security and Glean Partner to Secure Enterprise AI Agents
- 99 percent find AI agents where they aren’t expected
- China AI-Hardware-Heavy Funds Face Liquidation Wave as Changsheng Growth Select Triggers Wind-Up
- London’s $7 Billion AI boom reshapes global investment
- StudioX AI Brings Agent Interoperability to Air-Gapped Networks
- Criminal marketplace sells infostealer logs with AI-written profiles of their owners
- Companies Are Investing in AI, Despite Trusting It Less
- AI and 24-Hour Trading Test Rules Governing Currency Markets
- Can AI Review Your Investment Portfolio? What It Can and Can’t Tell You
Comments
Please log in to post a comment.