Anthropic has started watermarking text generated by Claude, a move to comply with EU’s AI Act that makes AI involvement detectable by other systems even though the watermark is invisible to readers. At the same time, LinkedIn is taking a much harder line on what it calls “AI slop,” reducing the visibility of suspected AI-generated content and giving users a way to flag content that appears overly reliant on AI.  While those are just two examples of changes, we expect their peers to follow suit with similar policies and policing.

The message for brands is clear: Content that feels generic, inauthentic or disconnected from a real human point of view and tone is increasingly at risk of being deprioritized — by platforms and by audiences.

At V2, we use AI as a thought partner. It helps us pressure-test ideas, find new angles, synthesize information and work more efficiently. But we believe the strongest communications still come from a human who knows the business and understands the audience. We advise our clients to take the same approach.

Best Practices for Protecting Against AI Watermarking and Slop:

• Protect executive voice and perspective. High-value communications such as thought leadership, internal communications, bylined articles and even social posts should reflect the actual ideas, experiences and voice of the individual. AI can help with structure, research or drafting, but the thinking should originate with the person whose name is attached to it and remain intact through the final draft. For example, start with an executive interview, voice memo, original notes or a substantive Q&A rather than asking an LLM to generate a point of view on the executive’s behalf.
• For high-value content, preserve the human trail. When authorship matters, it is increasingly valuable to be able to demonstrate how a piece came together. Communications teams can and should drive best practice around preserving interview notes, source materials, early outlines and version history to help show the progression from an executive’s ideas through drafting and editing. These add up to meaningful evidence of human authorship and judgment, should you ever be in a position to defend an idea or asset against AI claims.
• Be especially careful with content at scale. The pressure to produce more content faster can quickly lead to sameness. A higher volume of AI-assisted posts, articles and comments rarely create more impact, particularly as platforms and audiences get better at recognizing generic content. Ten posts that simply repackage conventional wisdom are unlikely to outperform two that contain an original observation, customer experience, proprietary data point or contrarian take. The goal should be stronger content, not simply more material.
• Expect AI detection — but understand its limitations. AI checkers, classifiers and watermark detection are becoming part of the content ecosystem, and brands should assume their content may be evaluated by them. But their results require context. A watermark or other signal showing that content passed through Claude or another LLM does not necessarily mean the AI authored the underlying ideas or entire piece; likewise, the absence of a detectable signal does not prove something was written entirely by a human. Treat these tools as signals about the content process, not definitive proof of authorship.
• Set clear expectations with employees and partners. AI use is already happening across organizations, whether there is a formal policy or not. Companies should be explicit about where AI can support the content process, where human review is required and which communications require primarily human authorship. For example, using AI to summarize background research or suggest alternate headlines is very different from asking it to generate an executive’s POV from scratch and publishing the result under that person’s name.
• Don’t overreact to an AI flag, but do have a plan. If a brand or executive post is flagged as appearing AI-generated, first assess what triggered the reaction: generic language, lack of a distinct point of view, excessive polish or structure, or actual overreliance on AI. A flag alone should not prompt a public defense or deletion of the content. If the underlying ideas and authorship are sound, retain the documentation that demonstrates the human process and use the feedback to strengthen future content. If the post genuinely does not reflect the executive’s voice or expertise, revise your approach rather than trying to “beat” the detector.

As AI becomes more embedded in the content process, we think the competitive advantage will come from using it well while preserving the thing audiences still value most: a real point of view — backed by real expertise, experience and human judgment.

Katelyn Holbrook is Chief Client Officer at V2 Communications, an integrated PR and communications firm headquartered in Boston. V2 helps technology companies improve visibility, strengthen reputation, and adapt to how audiences discover and evaluate brands in an AI-driven landscape.

About V2 Communications

V2 Communications is a top PR firm for tech companies and a leading integrated communications partner focused on B2B technology, AI, climate and energy, healthcare and high-growth brands. The firm delivers integrated communications programs across earned, owned, and paid media, helping companies build market leadership, increase visibility, and drive measurable business results.