Blog

AI and the Smart Program Director

By David Kidd, BPR

AI is no longer a toy for radio programmers. Used well, it is a force multiplier. Used badly, it is a fast way to lose trust, talent and possibly a regulator’s patience.

For programmers, the best use of AI is not to replace the craft of programming, but to remove the sludge around it. AI can summarise research, scan music logs, identify topic trends, draft promos, compare competitor clocks, transcribe shows, generate first drafts of social posts and help producers turn one good idea into ten usable executions.

That gives programmers more time for the two things AI still cannot do properly: taste and judgment.

The danger is pretending the machine is the talent.

For example, Australia has now drawn a clear line. Under the Commercial Radio Code of Practice 2026, in effect from 1 July 2026, commercial stations must disclose when a Regularly Scheduled Program or News Program that is hosted by a synthetic voice generated by artificial intelligence. However, the legislation does not apply in relation to “any other content that is broadcast during the applicable Regularly Scheduled Program or News Program, including weather, traffic, music and advertisements”.

Note: this legislation does not apply to the more than 450 Community Radio stations in Australia or the government funded broadcaster, the ABC. They can use AI where they see fit without disclosure.

The ACMA says this is the first time AI transparency has been included in an Australian broadcasting code. A synthetic voice doing a weather break, a character sketch or a production element may be harmless. The legislation is based upon the premise that a synthetic “announcer” presented as a real human without disclosure is different.

Australia is currently one of the first countries in the world to have a direct radio broadcasting rule requiring disclosure of synthetic presenters. Most other countries are regulating:

1. voice cloning and impersonation, and/or
2. deepfakes and political misinformation, rather than radio itself.

I suspect that over the next two to three years, regulators in the US, UK, Canada, Europe and New Zealand will move toward an Australian-style requirement: if listeners cannot reasonably tell that a presenter is AI, broadcasters will likely have to disclose it.

So the rule for programmers should be simple: use AI backstage aggressively, but on-air transparently.

AI should help with preparation, analysis, editing, repurposing and workflow. It should not secretly impersonate people, invent facts, manufacture fake listener calls, clone talent without consent or create “news” without human verification. Every station should have a basic AI checklist: disclose synthetic voices where the law requires it, keep humans responsible for ensuring news is factually correct to avoid legal risk, secure written consent for voice cloning, label AI-generated material where needed and never let speed outrank accuracy.

The winning programmer will not be the one who replaces the most people with AI. It will be the one who uses AI to make the station sharper, faster, sound more local and more prepared.  In radio, authenticity is still the secret ingredient. AI is just another piece of equipment in the rack.