Skip to content
You can now search across every topic, entity and event.What's new
Media's AI Pivot
17JUN

Fox buys Roku's data layer for $22bn

3 min read
09:26UTC

Fox Corporation agreed to buy Roku for about $22bn on 15 June, acquiring device-level data on 100m-plus streaming households. The same week, UK and EU regulators and 31 British publishers moved to wall off their own data from AI scrapers. Two halves of one fight over who owns the inputs that train and target media AI.

Key takeaway

Fox's $22bn Roku buy and publishers' legal walls are both bets on one fact: proprietary data, not model capability, is the competitive moat.

This briefing mapped
Economic
Regulatory
Legal
Competitive

Fox Corporation agreed to buy Roku for about $22bn on 15 June, gaining device-level viewing data from more than 100m streaming households and never once naming AI as the reason.

Sources profile:This story draws on neutral-leaning sources from United States
United States
LeftRight

Fox Corporation agreed on 15 June to buy Roku for $22 billion, gaining device-level viewing data from more than 100 million streaming households. The deal pays $160 per Roku share and is expected to generate $400 million in annual cost savings.

Fox One, Fox's new streaming service, launched on 11 June with AI-personalised dashboards for the World Cup. The Roku acquisition supplies the household data that makes those AI features commercially valuable to advertisers. 

The UK Competition and Markets Authority ordered Google on 3 June to give news publishers tools to withhold their content from AI search, the first binding order of its kind.

Sources profile:This story draws on neutral-leaning sources

The UK's Competition and Markets Authority ordered Google on 3 June to give news publishers tools to opt out of having their content used in AI Overviews, AI Mode and model training, and to add clear source links inside AI search results. The CMA cited measurable publisher traffic falls since AI Overviews launched.

No regulator had previously made such a requirement binding. The practical effect depends on a compliance Deadline the CMA has not yet published. 

Sources:TechXplore

The Movement for an Open Web signed 31 UK publisher sites to contracts charging AI crawlers £500 per scraped article, enforced through small-claims court rather than copyright law.

Sources profile:This story draws on neutral-leaning sources

The Movement for an Open Web launched Search-Only Contracts on 15 June, signed by 31 UK publisher sites including Trusted Reviews and road.cc. The scheme charges AI crawlers £500 per article copied without permission, enforced through England's online small-claims court for a £50 filing fee with no lawyer required.

Whether it works legally depends on whether English courts accept that an AI crawler is bound by a website's terms. No test case has been filed yet. 

The US Justice Department cleared Paramount Skydance's roughly $110bn purchase of Warner Bros. Discovery on 12 June, the first of several gates the deal still has to pass.

Sources profile:This story draws on neutral-leaning sources

The US Department of Justice cleared the merger of Paramount Skydance with Warner Bros. Discovery on 12 June, valuing the deal at roughly $110 billion. The combined company would hold CNN, HBO Max, the Warner Bros. studio and Paramount+.

Three clearances remain: the FCC must approve because the combined company would be nearly half foreign-owned, the EU has until 14 July to rule, and state attorneys general could still sue. No AI strategy has been announced for the merged entity. 

Sources:Deadline

France Televisions disclosed on 11 June the named AI vendors it ran live at Roland-Garros, the opposite of the unnamed generative layers US streamers will not discuss.

Sources profile:This story draws on neutral-leaning sources

France Televisions, the French public broadcasting group, disclosed on 11 June the full AI stack it used to produce Roland-Garros tennis coverage, naming Decart's Lucy 2.0 model for real-time video transformation, Gracia AI for volumetric 3D video and Topaz Labs for remastering.

Most broadcasters keep their AI vendor choices private. France Televisions' named-vendor disclosure sets a transparency standard that other European public broadcasters may face pressure to match, with the EU's AI content-marking rules taking effect on 2 August. 

The European Commission finalised its Code of Practice on marking AI-generated content on 10 June, completing a rulebook that applies from 2 August with no media company yet signed.

Sources profile:This story draws on neutral-leaning sources from Belgium
Belgium

The European Commission finalised rules on 10 June requiring AI-generated content to be labelled, applying from 2 August 2026. Companies can sign up voluntarily now; signing lets them show regulators they are meeting EU AI Act obligations. AI systems already deployed before 2 August have until December 2026 to comply.

Zero EU broadcasters or media companies had signed at finalisation on 10 June. the Commission is expected to make the rules mandatory if voluntary uptake stays low, following the same path as its 2018 disinformation code. 

Lionsgate extended its equity stake in generative-AI firm Runway on 12 June, choosing partnership over the acquire path Fox took and the build-and-rent path Netflix runs.

Sources profile:This story draws on neutral-leaning sources

Lionsgate extended its equity stake in Runway on 12 June to co-develop short episodic series from its film and TV library, building on a 2024 partnership that trained a custom Runway model on Lionsgate's roughly 20,000-title catalogue. Lionsgate injected no fresh cash and disclosed no terms for the extension.

Lionsgate vice-chair Michael Burns described the arrangement as an iterative process. Runway holds a $5.3bn valuation and counts Netflix and the BBC among its media clients, but Lionsgate has not named a distribution platform for the planned series. 

The New York State Assembly passed bill A11292 on 5 June requiring AI web crawlers to disclose their identity or face penalties of $15,000 a day.

Sources profile:This story draws on neutral-leaning sources

New York State's lower legislature passed a bill on 5 June requiring AI web crawlers to identify themselves truthfully or face fines of $15,000 per day. The bill still needs to pass the state Senate and be signed by the Governor to become law.

The bill matters because publishers cannot charge or block AI crawlers they cannot identify. Mandatory identification is the legal infrastructure that makes per-article billing schemes, including the UK's new publisher contracts , practically enforceable. 

Closing comments

Escalating. The EU Code of Practice's 2 August application date is the nearest hard deadline and the most likely first mover: a major European broadcaster signing before then will validate the Code's voluntary pathway; zero signatories by 31 July will accelerate the Commission's mandatory-conversion track, which replicated in 18-24 months the 2018 Disinformation Code's trajectory. The specific trigger to watch is the EU's ruling on Paramount-WBD by 14 July: if the Commission requires structural remedies on European streaming assets, it establishes an intervention precedent that extends beyond merger control to data-concentration concerns across the media-AI pipeline. The FCC's foreign-ownership ruling on Paramount's 49.5% figure carries parallel significance for US broadcast regulation. Neither outcome is decided; both are dated.

AI-assisted, human-edited under the editorial responsibility of Bannermedia Ltd. Reviewed by Ed Woodcock on 17 June 2026. Editorial standards.

Different Perspectives
Fox Corporation
Fox Corporation
Fox paid $22bn for Roku on 15 June without mentioning AI once in its announcement, acquiring device-level data from 100m-plus streaming households six months after FoxNXT began staffing a full-chain AI production team. The acquisition closes the data gap between the AI capability Fox built first and the first-party signal those products need to generate advertising premium.
UK data defenders: CMA and MOW publishers
UK data defenders: CMA and MOW publishers
The CMA issued the world's first binding order on 3 June requiring Google to let publishers opt out of AI scraping without losing search ranking, while 31 UK publishers signed MOW's Search-Only Contracts on 15 June charging crawlers £500 per article via small-claims court. Both tackle the same gap: blocking AI crawlers currently forfeits search visibility.
France Televisions
France Televisions
France Televisions named Decart, Gracia AI and Topaz Labs as its live Roland-Garros AI vendors on 11 June, the opposite posture to US streamers running undisclosed generative layers. The named disclosure positions France Televisions for Article 50 compliance before the EU AI Act's 2 August deadline and turns three startup vendors into European public-broadcaster reference customers.
European Commission and AI Board
European Commission and AI Board
The Commission finalised its Code of Practice on AI content marking on 10 June and opened sign-up with 46 days until the 2 August application date, but no EU broadcaster or media company had signed by close. Zero voluntary uptake replicates the 2018 Disinformation Code trajectory, which the Commission converted to mandatory standards within 18-24 months when adoption stalled.
AI labs and platform vendors
AI labs and platform vendors
OpenAI, Google and Anthropic are named as the targets of the MOW £500-per-article contracts, New York's A11292 crawler-disclosure requirement, and the CMA opt-out order in the same week. News Corp's anticipated $1.5bn Anthropic settlement (ID:3185) set the bilateral licensing benchmark; the legal architecture assembling this week is the infrastructure for publishers who cannot reach that settlement table.