Microsoft will no longer invest in third-party facial recognition firms, including AnyVision, after the Israeli company was accused of surveilling Palestinians
AnyVision was accused of helping the Israeli government surveil the West Bank — Microsoft says it will no longer invest …
Context & Ripple Effects
This exit closes a loop that opened in August 2019, when privacy activists argued that Microsoft's funding of AnyVision contradicted its public calls for facial recognition regulation. Microsoft responded by hiring former Attorney General Eric Holder to audit whether AnyVision met the ethical principles attached to its Series A (the Holder review).
The decision matters because Microsoft chose divestment over defense: rather than waiting on the review's findings, it is exiting third-party facial recognition investing altogether. The related coverage shows this was not a one-off — five years later the same logic reappears when Microsoft reportedly cut Israel's Unit 8200 off from Azure over mass surveillance findings.
First-order effects
- AnyVision loses Microsoft as a strategic investor at a moment when its flagship customer relationship is under public scrutiny, raising both its capital-raising difficulty and its reputational cost for future backers.
- Microsoft's venture arm must unwind or freeze positions in facial recognition startups, converting an equity portfolio into a liability it now has to manage rather than grow.
Second-order effects
- Other corporate investors in surveillance-adjacent AI startups face the same activist playbook — fund a company, get tied to its deployments — making ethical-use clauses a standard term in strategic rounds.
- Facial recognition vendors lose a class of deep-pocketed strategic capital and must lean more heavily on government contracts and financial-only investors, shifting their accountability structure away from tech-industry oversight.
Third-order effects
- The pattern holds across the corpus: what began as an investment screen hardened into platform-level enforcement, culminating in Microsoft terminating Unit 8200's Azure access and tightening human rights controls for national security work — suggesting cloud providers are becoming de facto arbiters of which state surveillance uses get compute.
- If equity stakes keep creating this kind of exposure, big tech may structurally retreat from owning pieces of sensitive-AI vendors, preferring arm's-length commercial relationships it can sever over ownership it cannot easily exit.
The trend: Major cloud companies are progressively walling themselves off from state surveillance applications — first by screening investments, eventually by cutting off the underlying compute — under sustained activist and employee pressure.