Lawmakers Demand FAA Pause AI Air Traffic Tool SMART

Key Takeaways

• Over two dozen House Democrats demand FAA pause SMART AI air traffic tool
• SMART launched around BWI, Dulles, and Reagan National airports last month
• Lawmakers call the rollout reckless and demand safety testing
• Study of 10 LLMs across 5 games finds strategic ability varies by opponent
• Multi-armed bandit algorithms outperform self-reflection methods in strategic AI tasks
• UK government undecided on dedicated AI laws vs. AI Security Institute powers
• OpenAI's admissions about AI agents were discussed in ABA Journal podcast on AI risks
• AI agents with human names and job titles are entering workplaces
• Security spending stays at 10-13% of IT budgets despite AI hype
• Experts debate whether AI could threaten cryptocurrency security systems

Democratic lawmakers demand FAA pause AI air traffic tool

More than two dozen Democratic lawmakers are calling on the Federal Aviation Administration to stop using an AI-powered air traffic control tool. The lawmakers are concerned about the tool's reliability and safety. They say the rollout was rushed and that frontline workers had little warning about the new system.

House Democrats push FAA to pause AI tool in DC airspace

Lawmakers are questioning the Federal Aviation Administration over its AI-powered tool called SMART. The tool launched last month around three D.C.-area airports including BWI, Dulles and Reagan National. It gives aviation specialists a view of 200 data streams to help reduce delays. Over two dozen House Democrats signed a letter calling the rollout reckless and demanding safety testing before the tool continues.

Study examines how LLMs make strategic decisions

New research looks at how large language models make strategic choices in games. The study tested 10 LLMs across 5 different games and found that a model's strategic ability changes depending on the game and the opponent. Lightweight methods like multi-armed bandit algorithms performed more reliably than self-reflection or query embedding approaches. The findings highlight challenges in picking the right AI model for strategic interactions.

UK AI laws and security institute powers still undecided

The UK government has not decided whether to create dedicated AI regulations or give the AI Security Institute legal powers for safety testing. A government official told lawmakers that civil servants are still working on an analysis of the wider AI regulatory framework. The decision was discussed at a parliamentary committee hearing on Tuesday evening.

Experts discuss whether AI could pose existential threat

Recent headlines about AI have raised alarming concerns about its potential risks. In a Legal Rebels Podcast episode, the ABA Journal's Victor Li discusses whether AI needs guardrails and how those guardrails should be designed. The conversation explores what lawyers can do to guide AI toward outcomes that benefit humanity. OpenAI's recent admissions about AI agents are among the topics covered.

AI Crypto Security Debate and Ethereum Bunker Mode

Experts are debating whether artificial intelligence could break the security systems that protect cryptocurrencies. Justin Drake proposed an Ethereum bunker mode as a possible response to these threats. Critics question if bunker mode would actually work in practice. The industry still disagrees on when AI security risks might become a real problem.

AI Agents Enter the Workplace with Human Names

Some employers are using advanced AI agents to handle complex multi-step tasks beyond simple chatbots. These AI agents are given human names, email addresses, and even job titles, being called colleagues. Employers must remember that AI agents cannot be held accountable when things go wrong. Companies should set clear policies so employees stay responsible for work quality and verify AI outputs carefully.

AI Jobs and the Future of Work

The author describes how AI now handles phone calls and tasks faster than humans can. Nobel economist Wassily Leontief warned that automation could leave most people without jobs. The author says wealth from AI is going to a tiny group of billionaires and trillionaires. He calls for heavy taxes on concentrated wealth and new job creation or universal basic income to address the problem.

Carrie Coon Criticizes AI and Supports Diversity in Hollywood

Gilded Age star Carrie Coon said AI is not generative art but an amplification machine that cannot create new ideas. She believes human artists will always be needed to make truly innovative art. Coon also spoke about the need for better representation of people of color, queer people, and immigrants in Hollywood. She praised the Gilded Age for showing the Black Diaspora in its time period.

AI Raises Stakes for Zero Trust Security

Security expert Kindervag says AI is mostly the same security game with more hype, since AI still operates within network protocols that can be controlled. He notes security spending has stayed between 10 and 13 percent of IT budgets for years. Zero Trust means treating every packet with zero trust and granting access based on what each user actually needs. Companies should treat new AI agents like interns with limited permissions that grow over time.

Sources

• Democratic lawmakers oppose AI tool used at D.C.-area airports
• House Democrats push FAA to pause AI tool deployed in DC airspace
• Choosing LLMs for Strategic Agent Interactions
• UK AI laws and security institute powers remain undecided, govt official says | MLex | Specialist news and analysis on legal risk and regulation
• Could artificial intelligence destroy us all?
• AI Raises Potential Crypto Security Threats, Fueling Debate Over Ethereum’s ‘Bunker Mode’
• AI Agents in the Workplace » CBIA
• What Will Everyone Do for a Living?
• 'Gilded Age' star Carrie Coon hits back at AI, talks diversity in Hollywood
• AI Raises the Stakes for Zero Trust