A lot moved this week. The EU made a significant call on AI compliance deadlines, Stanford dropped its annual benchmark, and the data on who’s actually getting hurt by AI displacement is sharper than the headlines usually give it credit for.
1. EU Trims the AI Act – and Bans AI-Generated Intimate Images
Source: EU Council | Euronews
On May 7, the EU struck a deal to simplify the AI Act – pushing major compliance deadlines back 16 months for most high-risk systems and extending SME-level exemptions to small mid-caps. The headline-grabber, though, is a new prohibition banning AI systems that generate non-consensual intimate images, effective December 2026. For small businesses, the deadline relief is real breathing room. For everyone, the deepfake prohibition is long overdue and should’ve been in the original text.
2. Stanford’s 2026 AI Index: The Gap Between Power and Transparency Is Widening
Source: Stanford HAI | IEEE Spectrum
Stanford’s annual AI Index landed this spring with some sobering data alongside the usual capability benchmarks. The Foundation Model Transparency Index dropped from an average score of 58 to 40 – meaning the most powerful models are less transparent about how they work than they were a year ago. Documented AI incidents rose from 233 to 362. Models are getting smarter and less legible at the same time. That’s not a hypothetical risk – it’s a documented pattern, and it’s the exact reason AI adoption needs people who can translate what’s actually happening inside these systems for the organizations using them.
3. Microsoft Report: AI Adoption Climbs, But the Global Divide Keeps Widening
Source: Microsoft On the Issues | Redmond Mag
Microsoft’s Q1 2026 AI Diffusion Report shows global AI usage climbed to 17.8% of the working-age population – but the gap between the Global North (27.5%) and Global South (15.4%) is widening, not closing. The UAE leads at 70.1%; the U.S. is at 31.3%. The communities with the least resources to navigate AI adoption are falling further behind, while the technology moves faster. This is the kind of structural inequity that responsible AI consulting should be talking about honestly, not just flagging in footnotes.
4. Young Workers Are Taking the Hardest Hit from AI Displacement
Source: Dallas Fed | DesignRush
New Dallas Fed data confirms what’s been building for a year: unemployment among workers aged 22-27 in AI-exposed roles has risen almost 3 percentage points since early 2025. Software developers aged 22-25 are down 20% from their 2022 employment peak. The mechanism isn’t mass layoffs – it’s fewer people entering those roles in the first place. Entry-level jobs are quietly disappearing, and the workers who’d normally be building their careers through them are finding the door closed. This deserves more honest conversation than it’s getting.
5. 68% of Small Businesses Use AI – and 77% Have No Policy Governing It
Source: Digital Applied | SBE Council
According to recent surveys, 68% of small businesses are using AI tools regularly – but 77% have no written policy covering how those tools are used, what data they touch, or what happens when they produce bad output. The average small business is running five different AI tools simultaneously without a clear framework for any of them. The adoption wave is real. The guardrails are lagging badly.
6. Suno Settled with Warner – Sony Fight Still Looms
Source: Music Business Worldwide | Chartlex Tracker
Suno closed its settlement with Warner Music Group in November 2025, striking a licensing deal and acquiring Songkick in the process. Sony’s case is still active – with a potential pivotal ruling expected this summer – and settlement talks reportedly stalled in April over how “walled” Suno’s platform would need to be. For anyone using AI music tools in creative or commercial work, this legal landscape is still unsettled. The outcome of the Sony case will shape what licensed AI music production actually looks like going forward.
7. Stability AI Releases Open-Weight Music Models That Generate 6-Minute Tracks
Source: Music Business Worldwide
Stability AI launched Stable Audio 3.0 – a family of four models trained on licensed data, three of them open-weight and free to use and build on. The big jump is track length: up to six minutes, more than double the previous ceiling. For producers and composers using AI as an ideation or drafting tool, open-weight models trained on licensed content start to look a lot more usable than proprietary black boxes.
8. Consumers Like AI Music – Until They Find Out It’s AI
Source: ProMarket
A study published this month found that listeners actually preferred AI-generated tracks in blind comparisons – but that preference reversed when they were told the music was AI-made. For human artists using AI as a collaborative tool rather than a replacement, there’s a real opportunity here to be transparent about the process without apologizing for it. Authenticity still has weight.
9. AI Tools for Content Teams Are Now Standard – Agentic AI Is Next
eMarketer reports AI is now the top tool content teams reach for to speed up production in 2026, with 51% of content professionals using it to accelerate creation, tagging, or organization. The shift happening now is from generative AI (help me make this) to agentic AI (do this for me). For small businesses, this means the gap between those with an AI-native content strategy and those without is widening faster than it was even six months ago.
10. AI Music Tool Revenue Up 651% Since 2023 – But Producers Are Staying in the Driver’s Seat
Source: IMS 2026 Report | Music Library Report
The International Music Summit’s 2026 data puts AI music tool revenue at $333 million annually – a 651% increase since 2023 – with 63 million monthly active users. But the survey of over 1,100 producers tells a more nuanced story: 60% use AI for ideation, 30% as a co-producer, and only 5% are delegating full production to AI. The tools are changing how music gets made – they’re not replacing the people making it.
That’s the week. AI is moving fast – faster than governance, faster than policy, faster than most organizations’ ability to manage it thoughtfully. That’s not a reason for panic. It’s a reason to keep building practical, people-centered AI frameworks that help real organizations navigate this without getting left behind or burned.
Freeland AI Collective helps small businesses and rural organizations adopt AI in ways that make sense – practical, ethical, and built around how real people work. More at freelandai.org.