Publishers that want to keep AI agents away from their articles should put the content behind a login, Bright Data senior director Jennifer Burns said in an interview with Cristian Dina at HumanX in Amsterdam. Anything a person can read without signing in counts as public and should remain open to AI as well, while a paywall qualifies as such a barrier. The position matters because Bright Data supplies web data infrastructure for AI models, agents and market research, and its tools shape how automated systems collect online content.
Burns position on public data and logins
Burns stated the company line as a clear distinction between public and non-public data, with no selective access at scale. If material is available to any visitor, she argued, it should also be available to AI, and attempts to draw other boundaries amount to gatekeeping. A publisher dissatisfied with collection by agents can close access and decide who is allowed inside. She added that the dispute over payment for content is real, but the market still has to resolve it, and that outcome is not for her company to decide.
Bright Data describes its business as infrastructure for collecting public web data, serving AI models, agents and research teams. The company reports 50,000 customers, up from 20,000 three and a half years ago when Burns joined after 20 years in local news journalism. It says it serves 14 of the top 20 LLMs and processes 100 million agent interactions per day. Staff grew from about 250 to around 600, and the company recently appointed a new CTO.
The interview took place for the TNW Podcast, which previously featured Atlassian Avani Prabhakar and Remote Job van der Voort. The same conflict is unfolding across the media market: Reddit said last week that it will require login, while the New York Times and three other US newspapers sued Brave over scraping. Bright Data itself has faced Meta and X in court over data collection, with the X case brought by that platform. Burns said courts have shown the company activity to be legal, and that it defended the case as it would any other.
What this means for companies using AI agents
For businesses that rely on agents for pricing, assortment or market monitoring, the login rule defines which sources remain technically and legally accessible. Open pages with catalogues, reviews and listings stay within the collection contour, while subscriber zones and paywalled publications drop out of automatic coverage. Small firms without data partnerships feel this divide first, since they depend on open sources, whereas large organizations can negotiate access or buy licensed feeds. The result is a split between freely readable web signals and closed content that requires contracts.
Demand for live context strengthens this division, according to Burns. In early 2026 the company moved into the context layer, meaning live data that tells an agent what is happening now, such as current prices. Without such updates, an answer from an LLM or agent becomes practically useless for commercial tasks. Robotics developers added a parallel request at the end of 2025, seeking public footage for training, including different glass sizes, shadows, home lighting and ways of placing a glass on a counter. Filming all variants from scratch would take too long when the material already exists online.
The marker to watch is whether more large platforms follow Reddit and introduce mandatory login, and how courts handle cases such as the newspapers action against Brave. A wider shift to closed access would shrink the pool of openly collectable data and raise the value of licensed agreements and live data feeds. For business, that would mean auditing which inputs behind agent decisions come from open pages and which require permission. Bright Data own advice points in the same direction: manage agents with a human in the loop rather than letting them run unsupervised.
