AI Rights Brief Daily

Proving the person behind the image

Today’s briefing moves from labelling synthetic files to identifying the people, performances and permissions behind them. Meta has formally committed to the EU Code of Practice on AI-generated-content transparency, aligning its platform labelling and detection work with the rules now applying across Europe.

Facebook is introducing free selfie-based verification intended to show that a real person operates a profile, although the badge does not confirm that everything the account publishes is authentic. Meta has also detailed new anti-tampering safeguards for the capture light on its AI glasses, raising practical questions about consent when photography becomes wearable and nearly hands-free.

In professional video production, Runway’s Act-Two can transfer a person’s movement, facial expression, speech and gestures onto another character, making performance permission distinct from permission to use the resulting character image. Runway users also face an August decision window as the company replaces its Unlimited subscription with a fixed-credit Max plan. Underneath these developments, C2PA’s current Content Credentials specification has expanded provenance into live video and more detailed editing histories. The connecting issue is no longer merely whether a file is marked as AI-generated. It is whether the workflow can prove who was captured, who performed, who consented and what happened to the media afterwards.

Meta is turning its EU transparency commitment into a platform-level promise

Meta has confirmed that it is signing the European Union’s Code of Practice on Transparency of AI-Generated Content. The company says the commitment aligns with its existing system for identifying and labelling generated content across its platforms, its research detector for Meta-made imagery and its participation in C2PA. Meta says it will work with the EU AI Office and other organisations on disclosures that remain interoperable and understandable rather than creating a confusing collection of competing labels.

Why visual creators should care

Meta is simultaneously a generator, publisher, advertising system and distribution platform. Its implementation choices will therefore affect how AI-generated and manipulated photographs, advertisements and videos are labelled at an enormous scale.

Practical takeaway

Do not assume Meta’s own label removes the creator’s disclosure responsibility. Preserve the original provenance and record the visible disclosure used in the post, advertisement or campaign. Check the published version because platform processing may display information differently from the source file.

What to watch next

Watch how Meta handles files created by third-party generators, whether C2PA information survives ordinary Instagram and Facebook uploads and whether Meta applies different labels to wholly generated media and AI-assisted edits.

Sources

Facebook’s new badge verifies a human account, not the truth of the images it publishes

Meta is introducing Facebook Verified, a free badge for eligible adult users who complete a short video-selfie check against their existing profile photographs. The badge will initially appear in areas including profiles, Marketplace, Dating and Groups, with Feed integration planned later. Meta says the badge means that a real person completed verification and meets relevant account standards; it does not represent an endorsement or guarantee that the account is trustworthy.

Why visual creators should care

For photographers, performers and visual artists, verified personhood could help distinguish an authentic account from an AI-driven impersonator. But the distinction needs to remain precise: a real person can still publish generated imagery, false claims or unlicensed material.

Practical takeaway

Treat the badge as evidence that a human controls the profile, not as evidence that an image is genuine or authorised. When licensing work or responding to an impersonation, continue checking the portfolio, source files, publication history and contact details.

What to watch next

Watch whether verification expands to professional-mode accounts and creators, whether impersonators can still build convincing verified composite identities and whether Meta creates a separate signal for verified creator ownership.

Sources

Meta is making its wearable camera stop working when the recording light is defeated

Meta says its AI glasses use a flashing white capture LED whenever photographs or video are being recorded. The camera is automatically disabled if the LED is covered, and Meta is updating the glasses to detect more sophisticated physical tampering or destruction of the light. The company also says it removes advertisements and listings for LED-tampering services and may take legal or account action against those offering them.

Why visual creators should care

Wearable cameras make photography less conspicuous than holding up a phone or conventional camera. A compulsory capture signal does not create consent by itself, but it gives nearby people a practical indication that recording is taking place.

Practical takeaway

Professional users should not rely solely on the LED when filming contributors, clients or members of the public. Use ordinary releases and verbal production protocols, and explain whether material will remain private, be uploaded to Meta AI or enter a commercial workflow.

What to watch next

Watch whether capture indicators become a legal or industry expectation for other AI wearables and whether platforms preserve information showing that footage originated from glasses rather than a conventional camera.

Sources

Runway can move a person’s performance into a character who looks nothing like them

Runway’s Act-Two workflow uses a driving-performance video and a separate character image or video. It can transfer facial expression, speech, movement and gestures from the performer to the character, and can subsequently apply a different generated voice. The resulting scene may retain little or none of the performer’s original visible appearance even though their physical and vocal performance shaped the output.

Why visual creators should care

Traditional releases often focus on the performer’s photographed face and body. Performance-transfer tools create a separate asset: the person’s acting, movement, timing, expression and speech can be used while their identity is visually replaced.

Practical takeaway

Obtain permission for performance transfer rather than relying only on a standard appearance release. Specify the character, project, intended contexts, voice treatment, permitted edits, reuse period and whether the performance may train or drive future generated characters.

What to watch next

Watch whether agency and performer agreements begin treating driving performances as reusable production assets and whether platforms expose durable records connecting a generated character to the human performance behind it.

Sources

Runway’s subscription change turns exploratory generation into a fixed production cost

Runway is replacing its Unlimited plan with Max on September 1. Existing Unlimited users retain their current service through August 31. Monthly subscribers move automatically at the same listed monthly price, while annual users may choose during August between migrating with a credit bonus or receiving a refund for the unused part of the subscription. Max provides 9,500 monthly credits and limited rollover but removes Unlimited’s relaxed-speed Explore Mode.

Why visual creators should care

The change affects how studios budget experimentation. An effectively open-ended exploratory workflow becomes a measurable credit cost, which may influence the number of iterations included in a quote and who carries the cost when a client repeatedly changes direction.

Practical takeaway

Recalculate the real generation cost of current Runway projects before migration. Distinguish included exploration from chargeable revisions in client estimates and decide whether annual migration or refund is appropriate during the August decision window.

What to watch next

Watch whether high-volume filmmakers reduce exploratory generations, pass credit costs to clients or move early development into lower-cost models before using Runway for approved production shots.

Sources

Content Credentials are moving from static files into live and streaming video

C2PA’s Content Credentials 2.3 specification adds support for live video, allowing provenance to operate in broadcast and streaming workflows. It also expands supported file types, provides more detailed descriptions of edits such as resizing, markup and redaction, improves links to trusted cloud-based information and strengthens validation and tamper detection. The update follows the launch of a conformance programme intended to make implementations behave consistently.

Why visual creators should care

Provenance is no longer limited to a photograph or completed video file. Live news, streams and event footage can begin carrying signed information about origin and production history while the material is being distributed.

Practical takeaway

Organisations adopting Content Credentials should decide who is authorised to sign live output and what editorial actions should appear in the history. A valid credential records claims and changes; it does not independently prove that the event shown is true or that every depicted person consented.

What to watch next

Watch adoption by broadcasters, livestream platforms and camera systems, and whether credentials remain available when live material is clipped, reposted, transcoded or converted into short-form social video.

Sources