The expertise needed to govern artificial intelligence is scarce and often concentrated within the very professional networks that fuel the industry's companies. The case of Matt Clifford makes this problem especially visible. Clifford, a central figure in British tech policy and chair of the Advanced Research and Invention Agency, announced his resignation after accepting a full-time role at Anthropic focused on government relations outside the United States.
According to The Guardian, the overlap between the two roles had sparked questions and parliamentary criticism over conflicts of interest. Clifford will leave ARIA in November, and safeguards are planned during the transition.
Expertise carries an institutional price
It is natural for governments to seek individuals who have a deep understanding of a complex industry. It is just as natural for those individuals to be attractive to companies willing to pay top dollar to understand institutions and regulators. The issue arises when the line between public advisory work and private interests becomes too porous.
One does not need to assume improper conduct to recognize the risk. Even the mere perception that those shaping the rules might soon work for those subject to them can erode trust in the policymaking process.
An issue set to grow
With AI becoming increasingly central to security, research, and the economy, states will need to build in-house expertise strong enough to avoid systematic reliance on industry. Otherwise, the revolving door will not be an exception, but a structural feature of tech governance.
