OpenAI gets the headlines. Google has the reach. But Anthropic may be the AI company publishers should watch most closely.

Their approach to media partnerships differs meaningfully from competitors. Understanding the difference could matter for how publishers navigate AI relationships.

The Landscape

AI companies are all negotiating with publishers now. The deals take different forms:

OpenAI has signed licensing agreements with major publishers including News Corp, AP, and various European outlets. The terms reportedly involve significant payments for content access.

Google has its Showcase program and various news initiatives, plus the ongoing Australian News Media Bargaining Code arrangements.

Microsoft through Copilot and Bing has its own publisher relationships, often bundled with broader Microsoft enterprise deals.

Perplexity has created a publisher program offering revenue sharing from ads displayed alongside cited content.

Anthropic has taken a notably different approach—one that’s received less attention but may matter more.

Anthropic’s Differentiation

Anthropic has positioned itself as more cautious about copyright and more respectful of publisher concerns than some competitors.

Several aspects of their approach stand out:

Opt-out respect. Unlike some AI companies that trained on publisher content without permission, Anthropic has been more responsive to robots.txt signals and publisher requests for exclusion.

Transparency emphasis. Their communication with publishers about how Claude uses and cites content has been more forthcoming than industry norm.

Attribution focus. Claude’s responses tend to include more explicit attribution to sources than some competing systems.

Conservative training claims. Anthropic has made fewer aggressive claims about fair use of copyrighted material for training.

This positioning appeals to publishers who’ve felt treated dismissively by other AI companies.

What the Deals Look Like

While specific terms are confidential, I’ve gathered information from multiple sources about the general structure of Anthropic partnerships:

Not primarily licensing. Unlike OpenAI’s approach, Anthropic deals aren’t primarily about paying for content access. The relationships are more collaborative.

Integration focus. Partnerships emphasize how publishers can use Claude rather than just how Anthropic can use publisher content. Tools, APIs, preferential access.

Research collaboration. Some deals include joint research on responsible AI use in journalism.

Policy development. Publishers involved in partnerships influence how Claude handles news content.

The value proposition is different: not “we’ll pay you for your content” but “we’ll help you use AI effectively while respecting your concerns.”

Why This Matters

For publishers considering AI partnerships, Anthropic’s approach has implications:

Leverage in negotiations. Anthropic’s positioning gives publishers more leverage. An AI company that prioritizes publisher relationships can be held to that standard.

Integration opportunities. The emphasis on helping publishers use AI means potential access to sophisticated tools and preferential technical support.

Influence on development. Publishers in Anthropic’s orbit may have more voice in how the technology evolves.

Reputation association. Partnering with an AI company perceived as responsible has different reputational implications than partnering with more aggressive players.

The Skeptical View

Not everyone is convinced Anthropic’s approach represents meaningful difference.

Critics point out:

Economic constraints. Anthropic may simply have less capital than OpenAI or Google, making expensive licensing deals impractical.

Training already done. Current models were trained on content that included publisher material. Responsibility now doesn’t undo past use.

Competition could change approach. As Anthropic competes more aggressively, their caution may diminish.

Words versus contracts. Public positioning is one thing; binding legal commitments are another.

These critiques have merit. Publishers should evaluate actual terms and commitments, not just positioning.

Practical Implications

If you’re a publisher considering AI partnerships, here’s what I’d suggest:

Engage with multiple companies. Don’t assume any single AI company is definitively better. Negotiate with several and compare terms.

Look beyond payments. Licensing fees matter, but so do integration opportunities, policy influence, and technical support.

Get specifics. Vague commitments to “respect publishers” are worthless. Push for contractual terms.

Consider reputation. Which AI companies you partner with publicly signals your values. Choose partners whose reputation you’re comfortable associating with.

Think long-term. These relationships will evolve as AI technology and the regulatory environment change. Consider flexibility and exit terms.

The Australian Context

For Australian publishers, the AI partnership landscape intersects with existing regulatory frameworks.

The News Media Bargaining Code established precedent for platform-publisher negotiations. Whether similar frameworks will apply to AI companies remains unclear, but the possibility shapes negotiating dynamics.

ACCC has expressed interest in AI’s impact on media. Regulatory attention could create new leverage or obligations.

Publishers negotiating AI deals should consider how agreements might interact with potential future regulation.

Working with AI Companies

Beyond formal partnerships, publishers interact with AI companies through their use of AI tools.

Many newsrooms now use Claude, ChatGPT, or other AI assistants for research, drafting, analysis, and production. These usage relationships differ from content partnerships but matter too.

The AI companies are learning from how journalists use their tools. This informs development priorities and capability investment.

For larger implementations, some publishers work with intermediaries. Consultancies like team400.ai help publishers evaluate options, negotiate terms, and implement solutions across multiple AI platforms.

What I’m Watching

Several developments will shape how this plays out:

Copyright cases. Pending litigation could establish rules about fair use of publisher content for AI training. Outcomes will affect everyone.

Regulation. EU, US, and potentially Australian regulation of AI will constrain what’s possible in these relationships.

Competitive dynamics. As AI competition intensifies, will companies like Anthropic maintain their relatively publisher-friendly positioning?

Economic pressure. Will publishers’ financial constraints force acceptance of terms they’d otherwise reject?

Technology evolution. How AI capabilities develop affects what partnerships make sense.

Recommendations

For publishers navigating this space:

Engage now. The AI relationship landscape is forming. Publishers who engage early have more influence than those who wait.

Maintain relationships across companies. Don’t lock in exclusively with any single AI player.

Prioritize practical value. Integration and capability access may matter more than content licensing payments.

Document everything. These relationships will be analyzed and potentially contested. Clear records matter.

Seek expertise. AI partnership negotiation is specialized. Bringing in advisors—whether legal, technical, or strategic—can help. A Sydney-based firm we’ve spoken with understands both the technology and the media industry context.

The AI-publisher relationship is still forming. How publishers navigate it now will shape their position for years to come.

---

Are you involved in AI partnership negotiations? I’m interested in hearing about publisher experiences—anonymously if needed.