The AI landscape doesn't move in one direction — it lurches. Some techniques leap from experiment to table stakes in a single quarter; others stall against regulatory walls, technical ceilings, or organisational inertia that no amount of hype can dislodge. Knowing which is which is the hard part. The State of Play cuts through the noise with a rigorously maintained index of AI techniques across every major business domain — classified by maturity, evidenced by real-world adoption, and updated daily so you always know where you stand relative to the field. Stop guessing. Start knowing.
A daily newsletter distilling the past two weeks of movement in a domain or two — delivered to your inbox while the index updates in the background.
Each dot marks the weighted maturity of practices within a domain — hover for a brief summary, click for more detail
AI for generating and editing images, video, audio, 3D assets, and cross-media content. Mostly leading-edge with rapid advancement — image generation, music composition, and voice synthesis are approaching good practice. Video generation and 3D asset creation are progressing fast but quality and controllability gaps persist. The most active domain by momentum: over half the practices are advancing.
The headline: Making creative work with AI is now cheap and easy. Getting audiences, courts, and your own creative staff to accept it is the expensive part — and each of those three is now measurable in revenue.
Almost every organization is already using AI somewhere in creative production, whether or not anyone signed off on it: US Federal Reserve data puts generative AI use among arts, design, entertainment and media workers at 51 percent, up six points in a single quarter. The companies genuinely ahead picked the boring use cases — sports highlight reels, foreign-language dubbing, e-commerce product backgrounds, internal training video, background music — where savings are real because the output is functional and nobody's name is on it. The ones struggling pointed AI at consumer-facing brand work, where 78 percent of consumers now say AI makes advertising less authentic and 63 percent say it makes them less likely to buy. The window that is closing is not a technology window. It is the window to get consent records, content labeling, and disclosure practice in place before a regulator or a plaintiff asks for them.
Two AI transparency laws went live on the same day. The EU AI Act's Article 50 and California's AI Transparency Act both became enforceable on August 2, requiring machine-readable labels on AI-generated media, with EU fines up to €15M or 3 percent of global revenue. Midjourney was documented as non-compliant on day one. If you publish AI-generated images, audio, or video into the EU or California, the labeling obligation is live now, not pending.
The labeling technology those laws depend on turned out to be trivially removable. An open-source tool with 4,500 GitHub stars strips both Google's invisible watermarks and C2PA credentials — the industry provenance standard — from images and video in one step. Treat labeling as a compliance obligation you must meet, not a control that will stop misuse of your content.
Courts in two countries reached opposite conclusions within a week. A Munich court ruled on July 31 that the music generator Suno infringed copyright under German and US law, making licensing legally mandatory in Europe; days earlier the Delhi High Court ruled in OpenAI's favor on training data. Over 130 copyright cases now run against AI companies alongside 100-plus signed licensing deals, so ask vendors which jurisdictions their training data is actually licensed for.
Two major image tools were pulled within days of launch. Google withdrew Nano Banana 2 from Google Earth after 24 hours when it produced fabricated refugee camps, nuclear plants, and bomb craters despite carrying invisible watermarks; Meta rolled back Muse Image after three days over consent. Guardrails — the safety rules meant to stop AI doing the wrong thing — are failing at the biggest vendors under real-world use, which argues for keeping a person reviewing output before it ships.
The productivity claim got independently tested and did not hold. A survey of over 2,000 B2B marketers found 75 percent spending three or more hours a week fixing AI output, and only 4 percent reporting a net time saving. Before expanding a creative AI rollout, measure rework hours, not generation volume.
EU and California enforcement will produce its first penalties. Both regimes are operative and neither has a grace period. Inventory every AI-generated asset your organization publishes, confirm each carries provenance metadata at export, and document who approved it — that record is the defense.
Voice and likeness liability is consolidating fast. Nine federal class actions were filed this month against Adobe, Amazon, Apple, ElevenLabs, Google, Meta, Microsoft, NVIDIA and Samsung over voiceprint collection without consent, while Japan and Mexico each moved toward treating voice as a protected personal right. If you use synthetic voices or digital replicas of real people, get written, dated consent on file for every voice and face before the next campaign.
Platform monetization rules are tightening around volume-produced AI content. YouTube's July policy bars generic, repetitive AI content from a Partner Program controlling roughly $40.4bn in annual advertising, and TikTok's disclosure toggle carries a 30–40 percent reach penalty. If your content plan assumes AI lets you publish more for less, model the distribution penalty into the business case now.
Disclosure does not buy back trust. Consumers overwhelmingly say they want AI content labeled — and then engage with it less once it is. Real user-generated content holds an 83 percent brand-trust rating against 71 percent for AI equivalents, and AI video returns the lowest ROI of any AI application McKinsey measures, at 1.1x against 3.2x for text. Transparency is right and it will cost you reach.
Your creative team is adopting these tools without believing in them. A survey of 882 creative professionals found 86 percent using AI, 10 percent believing it is good for the industry, and 69 percent reporting burnout. Compulsory use plus low conviction plus exhaustion predicts attrition more reliably than it predicts productivity.
Purely AI-generated work cannot be owned. US and EU rules both hold that content with no human authorship gets no copyright protection, so the assets you generate most cheaply are the ones you can least defend against a competitor copying them. Hybrid human-plus-AI workflows are the legally sound route, and they erode a meaningful part of the cost saving.
Go deeper: the full Creative & Generative Media briefing — the longer analytical write-up, plus every practice we track in this domain with its maturity rating, the tools to consider, and the evidence behind our assessment.