Meta (META) and Microsoft (MSFT) work to wean staff off Anthropic's Claude, reports The Information

Vendor-concentration stories of this kind have become a recurring feature of the enterprise AI cycle: large buyers diversify or restrict external model providers after dependency, security, or data-leakage concerns surface, and the pattern has historically read as a portfolio-of-suppliers shift rather than a repudiation of any one vendor. The case distinction that matters is whether this is an internal-compliance tightening, which is routine at firms with competing model ambitions, or a broader commercial rupture affecting Anthropic's revenue trajectory and enterprise standing. Both named companies have their own frontier-model programmes, so prior form runs toward favouring in-house capability over third-party tools where the two overlap, a dynamic that has played out repeatedly as hyperscalers build parallel stacks. The read-across for Anthropic is less immediate revenue loss than signal value: enterprise adoption of external models has tended to be sticky once embedded in workflows, and forced migrations are typically slow and partial. What follows in episodes of this kind is clarification from the companies involved, a check on whether restrictions are usage policy or contractual, and any knock-on commentary from Anthropic or its cloud partners. As a press report rather than a disclosed filing, the item sits in the unconfirmed tier until either company addresses it.