OpenAI and Minneapolis Fed Warn of AI Accountability Risks

An AI-assisted cyberattack targeted a key European Union institution in October, the first such incident reported by the 27-member bloc. Some EU digital tools remained offline or were affected, but officials did not identify them or name those responsible. The incident highlights how AI tools can be used to find vulnerabilities as well as misused in attacks.

Researchers warn that SpaceX’s proposed network of one million orbital data centers could carry significant environmental costs. A study estimates rocket-launch emissions could rise more than 150-fold from 2025 levels, while returning spacecraft could produce 70 to 150 times more nitrogen oxides. Burning satellites could release about 200,000 metric tons of material into the upper atmosphere annually—more than 200 times the 2024 amount. The estimates depend on assumptions because SpaceX has not provided basic details about its rockets, satellites or materials; the authors call for more information and oversight.

AI is also moving into high-stakes decisions, prompting calls for human accountability. OpenAI representative Ball argues that people, not AI agents, must remain responsible for economic decisions, with audit trails, limited permissions and the ability to revoke an agent’s authority. He urges caution about agents owning property, forming companies or operating in the economy without supervision. In a separate warning, Minneapolis Fed President Neel Kashkari says concentrated AI investment has helped support U.S. growth, but growth could slow if the technology fails to lift productivity or deliver expected returns. He remains optimistic while noting that some investments may be misguided.

Questions about responsibility extend to military and financial uses. On November 17 in Sofia, Cardinal Michael Czerny presented Pope Leo XIV’s encyclical Magnifica humanitas and said lethal-force decisions must remain in human hands, with identifiable makers and a clear chain of responsibility. He called for time for human judgment and protections for civilians and essential infrastructure. At a World Federation of Exchanges session in Mumbai, SEBI Whole-Time Member KVR Murty warned that adaptive AI trading agents could be hard to trace and could amplify volatility if investors act on the same models and data. SEBI has established a verification agency to review investment and algorithmic-trading claims; Murty also pointed to AI’s potential in surveillance, compliance and fraud detection, and urged market institutions to invest in research and innovation labs.

AI is increasingly used in frontline services and personal life. 911 dispatch centers use it to help assess calls and direct resources; one system assessed a caller’s symptoms and sent an ambulance. Dispatchers nevertheless raise concerns about accuracy, human error and opaque decisions, and are expected to continue relying on their training. ChatGPT and other chatbots are also becoming sources of relationship advice and emotional support. Kantar found 54% of consumers across surveyed markets had used AI for an emotional or mental-wellbeing purpose, while a Global Dialogues study found 48.2% of participants in 70 countries used it for emotional support or personal issues at least weekly. Experts say AI can help prepare for difficult conversations but cannot resolve relationship conflicts. Writer Lindsey Hall’s discovery of her boyfriend’s private ChatGPT conversations contributed to their breakup.

In business software, 12th Tech is developing AI features for its Crescendo platform, beginning with Crescendo Hub. Planned tools include AI-assisted photo review and intelligent audit recaps and summaries to help users check collateral information, spot items needing attention and identify risks in collateral verification and wholesale finance workflows. The company says it will share more as it expands AI capabilities.

Courts are seeing more signs of AI use, too. A Stanford study found indications of AI in more than a third of civil lawsuits filed this year in Santa Clara and San Francisco counties, excluding debt collection cases. In Santa Clara, the share of complaints by self-represented people showing signs of AI rose from 19% in 2024 to 72% in 2026; about one-third of attorney-filed complaints also showed signs. Researchers say AI could widen access to legal knowledge but may add strain to crowded courts and introduce false citations or facts. The study’s co-authors caution that AI use alone does not establish a document’s quality or improve a person’s chances of winning.

Key Takeaways

  • The EU confirmed its first reported AI-assisted cyberattack, targeting a key institution in October; affected tools and the perpetrators remain unidentified.
  • A study estimates SpaceX’s proposed one-million-satellite data-center plan could multiply launch emissions more than 150 times against 2025 levels and release about 200,000 metric tons of material into the upper atmosphere each year; the estimates rely on undisclosed project details.
  • OpenAI representative Ball calls for human accountability over AI agents, including audit trails, restricted permissions and the ability to revoke authority.
  • Minneapolis Fed President Neel Kashkari warns that concentrated AI investment could weigh on U.S. growth if productivity and investor returns fail to meet expectations.
  • Cardinal Michael Czerny says humans must retain control of lethal-force decisions, with clear responsibility, time for judgment and safeguards for civilians.
  • SEBI’s KVR Murty warns adaptive AI trading agents could be difficult to trace and that similar models and data could drive synchronized trades and volatility; SEBI has created a verification agency.
  • AI tools assist 911 dispatchers with call assessment and resource decisions, but concerns remain about accuracy, human error and transparency.
  • Surveys report widespread AI use for emotional support: 54% of consumers in Kantar’s surveyed markets had used it for an emotional or mental-wellbeing purpose, and 48.2% of Global Dialogues participants across 70 countries used it weekly for emotional or personal issues.
  • 12th Tech plans AI photo review and audit summaries for Crescendo Hub, aimed at collateral verification and wholesale finance workflows.
  • A Stanford study found signs of AI in more than a third of eligible civil lawsuits in two Bay Area counties; in Santa Clara, the share among self-represented complaints rose from 19% in 2024 to 72% in 2026.

AI powered cyberattack targeted the European Union in October

An official in Brussels said a cyberattack using artificial intelligence targeted the European Union in October. Some of the bloc’s digital tools remained offline, but the official did not say which tools were affected or who was behind the attack. This was the first time the 27 member bloc reported a cyberattack using an AI tool. The official warned that AI assisted tools can be used to find vulnerabilities and can also be misused.

AI generated cyberattack hit a key European Union institution

Cybersecurity experts had been dealing with an AI generated attack on a key European Union institution for almost a month when an official confirmed it on Thursday. Some of the bloc’s digital tools were affected, but the official did not identify them or say who carried out the attack. It was the first reported cyberattack using an AI tool against the 27 member bloc.

Study warns SpaceX AI satellite plans could sharply increase pollution

A study estimates that SpaceX’s proposed million orbital data centers could increase rocket launch emissions by more than 150 times compared with 2025 levels. Spacecraft returning to Earth could produce 70 to 150 times more nitrogen oxides, gases that can help damage the ozone layer. The researchers also estimate that burning satellites could release about 200,000 metric tons of material into the upper atmosphere each year, more than 200 times the amount from satellites and other objects in 2024. The estimates rely on assumptions because SpaceX has not shared basic details about its rockets, satellites, or the materials they would use. The study’s authors say more information and oversight are needed to assess the possible harm.

AI tools help 911 dispatchers assess calls and send help

Many 911 dispatch centers are using artificial intelligence to help identify the nature of calls and send the right resources. In one example, an AI system assessed a caller’s symptoms and sent an ambulance to a medical emergency, helping the caller receive care in time. Some dispatchers worry about AI accuracy, human error, and a lack of transparency in how systems make decisions. The article says use of the technology is expected to grow, while dispatchers will continue to rely on their training and experience.

Cardinal Czerny says humans must answer for AI used in war

Cardinal Michael Czerny said in Sofia on November 17 that humans must remain directly responsible for violence and destruction caused by their technologies. Presenting Pope Leo XIV’s encyclical Magnifica humanitas at a gathering hosted with the Bulgarian Academy of Sciences and the Apostolic Nunciature in Bulgaria, he said decisions about lethal force must stay in human hands and their makers must be identifiable. Czerny said Pope Leo considers the “just war” theory outdated, while upholding the right to self defense in the strictest sense and pointing to dialogue, diplomacy, and forgiveness as ways to resolve conflict. He outlined three safeguards for AI in warfare: a clear chain of responsibility, enough time for human judgment, and protection for civilians and the infrastructure they need. Czerny also called for people to reduce hostility in their everyday choices and actions.

OpenAI’s Ball says humans must stay accountable for AI agents

An OpenAI representative identified as Ball argues that people should remain responsible for AI agents instead of handing control of the economy to machines. He calls for audit trails, limited permissions and the ability to revoke an agent’s authority. Ball also urges caution around agents owning property, forming companies or acting in the economy without supervision. The article notes that hidden reasoning and silent testing before wide releases make oversight important, while acknowledging that the economy is decentralized and the timing of change is uncertain.

Neel Kashkari warns major AI investment could slow US growth

Minneapolis Fed President Neel Kashkari said large investments in artificial intelligence create concentration risks for the U.S. economy. AI investment has helped drive U.S. economic growth in recent years, but growth could slow sharply if AI fails to raise productivity or deliver the returns investors expect. Kashkari said it is still unclear whether AI will produce stronger productivity growth or other returns. He remains optimistic but said some investments could be misguided, and noted that AI infrastructure is not the only factor driving inflation.

AI is becoming a confidant in romantic relationships

People are increasingly turning to AI chatbots for relationship advice, help with difficult conversations and emotional support. Kantar research found that 54% of consumers across the markets surveyed had used AI for at least one emotional or mental wellbeing purpose, while a Global Dialogues study found that 48.2% of participants across 70 countries used AI for emotional support or personal issues at least weekly. Therapist Georgina Vass says AI can help people prepare for hard conversations but warns that it cannot resolve relationship conflicts for them. Coach Ruchi Ruuh says AI offers convenience, but human relationships help people learn through resistance and practice. The article also describes how writer Lindsey Hall’s discovery of her boyfriend’s private ChatGPT conversations contributed to their breakup.

12th Tech adds AI tools to its Crescendo platform

12th Tech is developing AI features for its Crescendo platform, with initial tools planned for Crescendo Hub. These include AI-assisted photo review and intelligent audit recaps and summaries to help users review collateral information and spot items needing attention. The company is designing the features for collateral verification and wholesale finance workflows. Director of Software Development Jeremy Walker and Vice President of Business Development Casey Holyk said the tools aim to make existing information more useful and help lenders work efficiently and identify risks. 12th Tech plans to share more details as it adds AI capabilities across Crescendo.

SEBI’s KVR Murty calls for accountability rules for AI trading agents

Securities and Exchange Board of India Whole-Time Member KVR Murty warned that AI trading agents that change their strategies as they learn could create new regulatory challenges. Speaking at a World Federation of Exchanges session in Mumbai, he said regulators need ways to ensure agents can be traced and that an accountable institution can explain their decisions. Murty also warned that many investors using agents based on the same models and data could make similar trades at once and increase market volatility. SEBI has created a verification agency to review risk and return claims from investment advisers, research analysts and algorithmic trading providers, and sees possible uses for AI in surveillance, compliance and fraud detection. Murty also urged market infrastructure institutions to invest in research and development, including dedicated innovation labs, and flagged cloud, data and model providers as sources of systemic risk.

AI Use Rises in Bay Area Lawsuits, Especially Among People Without Lawyers

A Stanford University study found signs of AI use in more than a third of civil lawsuits filed this year in Santa Clara and San Francisco counties, excluding debt collection cases. In Santa Clara, the share of complaints filed by people representing themselves that showed signs of AI rose from 19% in 2024 to 72% in 2026. About one-third of complaints filed by attorneys in the county also showed signs of AI. Researchers said AI may help people gain access to legal knowledge, but it could also add pressure to crowded courts and lead to false citations or facts. Stanford law professor and study co-author Dan Ho said AI might improve some complaints, while Riana Pfefferkorn said using AI does not by itself show that a legal document is good or that it will help someone win.

Sources

NOTE:

This news brief was generated using AI technology (including, but not limited to, Google Gemini API, Llama, Grok, and Mistral) from aggregated news articles, with minimal to no human editing/review. It is provided for informational purposes only and may contain inaccuracies or biases. This is not financial, investment, or professional advice. If you have any questions or concerns, please verify all information with the linked original articles in the Sources section below.

AI-assisted cyberattacks EU cybersecurity AI agents AI governance human accountability AI safety space-based data centers SpaceX environmental impact of AI AI investment AI productivity autonomous weapons AI trading financial regulation AI in emergency response AI emotional support AI in business software collateral verification AI in legal services AI-generated legal filings

Comments

Loading...