AI News: The 'AI Bowl' lands with a thud as viewers reject tech's biggest ad blitz

Security & Privacy

40% of Employee AI Use Now Involves Sensitive Corporate Data

New research from Cyberhaven reveals that nearly 40% of all employee interactions with AI tools involve sensitive corporate data - and most of it flows through unsanctioned applications that IT departments don’t control.

The scope of the problem is striking. More than four out of five of the top 100 GenAI SaaS apps are classified as medium, high, or critical risk. The number of employees using generative AI applications has tripled, while data policy violations have doubled. The average organization now experiences 223 AI-related data security incidents per month.

Employees are choosing niche AI tools that offer specialized workflows and require no prompting expertise, bypassing official channels in the process. Microsoft’s own Data Security Index report, published in January, paints a similar picture: only one in three organizations has established comprehensive AI governance frameworks, creating what amounts to a permanent gap between how fast workers adopt AI tools and how fast security teams can evaluate them.

The pattern is familiar from the early days of cloud computing and shadow IT, but the stakes are higher. When employees paste proprietary code, financial data, or customer information into third-party AI tools, that data becomes training material for models they don’t control. The security community’s term for this - “shadow AI” - undersells the problem. This isn’t rogue software; it’s a structural failure to match AI adoption speed with data governance.

Sources: TechNewsWorld, Microsoft Security Blog

Regulation & Policy

The Federal-State AI Regulation Standoff Enters Its Critical Month

February 2026 is shaping up to be a decisive month in the fight over who regulates AI in America. President Trump’s December executive order directed the Attorney General to establish an AI Litigation Task Force to challenge state AI laws, with a March deadline for action. Meanwhile, states keep legislating.

Colorado’s AI Act - the first comprehensive state AI law in the country - was delayed from February 1 to June 30 after contentious negotiations collapsed during a special session. Governor Polis signed a bill that literally did nothing but find-and-replace every instance of “February 1, 2026” with “June 30, 2026,” kicking the can five months down the road. The law itself remains unchanged: when it takes effect, it will still require transparency and accountability for high-risk AI systems used in employment, healthcare, and education decisions.

California’s Transparency in Frontier AI Act took effect January 1 and is already operational, requiring developers of powerful AI models to implement safety protocols and report critical safety incidents. Texas’s Responsible AI Governance Act went live the same day. New York now has two AI-related bills in play - the data center moratorium and the FAIR News Act - with more likely coming.

The federal government’s strategy is to declare state laws incompatible with national policy and sue them out of existence. Whether that works depends on courts that haven’t yet been asked to weigh in. In the meantime, twenty states have AI-specific laws either passed or in development, creating a patchwork that companies have to navigate regardless of what Washington prefers.

Sources: Reed Smith, ETC Journal

Quick Hits

Anthropic’s anti-ad stance may age poorly: Adweek spoke to advertising industry creatives who noted that Anthropic’s “no ads in AI” Super Bowl campaign risks looking hypocritical if the company eventually needs ad revenue to fund its massive compute costs. Several pointed out that promising “never” in a capital-intensive industry is a bet against economics. Adweek
Seahawks win the actual game: In case anyone forgot there was a football game happening between the AI ads, the Seattle Seahawks defeated the New England Patriots at Levi’s Stadium in Santa Clara. CNBC
AI regulation at 20 and counting: Twenty US states now have AI-specific laws either enacted or in legislative development, creating a compliance landscape that grows more complex by the month - especially with the federal government actively trying to preempt state action. Drata

Worth Watching

The Super Bowl AI ad debacle deserves more attention than the industry will probably give it. When you spend an estimated $7 million per 30-second spot to reach 120 million viewers and your campaign lands in the bottom 3% of likeability, the problem isn’t creative execution - it’s that consumers don’t want what you’re selling, at least not the way you’re selling it. The AI industry has spent the past four years talking to investors, developers, and each other. The Super Bowl was its first real attempt to talk to everyone else, and the response was confusion, indifference, and hostility. That gap between industry enthusiasm and public reception is the single biggest risk to AI adoption that nobody in Silicon Valley seems willing to address.

The New York FAIR News Act also bears watching. It’s the most aggressive attempt yet to regulate AI in journalism, and its labor protection provisions - explicitly barring AI from displacing journalists or cutting wages - go well beyond transparency. If it passes, expect similar bills in other states. And expect the Trump administration’s AI Litigation Task Force to add it to its target list, setting up a First Amendment vs. state labor rights collision that could define how AI intersects with media for years.