AI Anchors Aivan and Aira Debut on Radyo Pilipinas, Promising Speed and Neutrality

AI bot-casters debut on government radio: What public communicators need to know

Government radio just put synthetic anchors on air. The Presidential Broadcast Service-Bureau of Broadcast Services (PBS-BBS) launched Aivan and Aira, AI-powered "reporters," during AI Talks on Radyo Pilipinas as part of its 78th anniversary.

PBS-BBS leaders say the goal is reach, speed, and consistency. "Actually, it's still us who write the news. They just do the execution," said Director General Fernando Amparo Sanga. "Since this is AI-generated, it is faster, it enhances the content."

The shift follows a memorandum of understanding signed on August 7 with voice veteran Pocholo Gonzales to run AI Talks with The VoiceMaster, described as the country's first AI-driven radio show. "Hindi lang ito basta programa… ito ay isang rebolusyon sa paraan ng paghahatid ng balita at kaalaman," he said. "Balitang AI, the world's first AI-animated news reporter, shows that we can be trailblazers, not just followers."

Why it matters for government teams

• 24/7 coverage: Push urgent advisories, disaster updates, and public service reminders on short notice.
• Consistency: Standardized delivery across regions and programs.
• Cost and capacity: Scale output without adding shifts, while keeping editorial control in-house.
• Multilingual reach: Faster localized versions if models support local languages and dialects.

Watchouts you need to address

• Bias and accountability: Bias lives in scripts, sources, and editing-not just voices. Keep human editorial responsibility explicit.
• Accuracy: AI can mispronounce names, abbreviations, and places. Build a pronunciation and style library.
• Misinformation: Require pre-broadcast fact checks and post-broadcast corrections logs.
• Public trust: Disclose AI use clearly to avoid confusion and maintain credibility.
• Data privacy: Ensure voice and content pipelines comply with local privacy regulations.

Implementation playbook

• Define use cases: Alerts, headlines, routine bulletins, multilingual rebroadcasts. Exclude sensitive or developing stories from automation.
• Set editorial rules: Source standards, fact-check steps, escalation triggers, and red lines (e.g., no AI for condolence or high-risk statements).
• Pick the stack: TTS engine, script editor, pronunciation dictionary, version control, and audit logging.
• Pilot with guardrails: Limited time slot, predefined topics, and parallel human broadcast for comparison.
• Quality loop: Daily air-checks, pronunciation fixes, audience feedback channel, and weekly error review.
• Legal and compliance: Privacy review, IP and voice rights, accessibility checks, and records retention.
• Incident response: Rollback plan to human anchors, correction workflow, and public notice template.
• Training: Upskill producers and editors on prompt writing, verification, and voice tools.

KPIs to track

• Time-to-air: Minutes from script finalization to broadcast.
• Error rate: Mispronunciations, factual corrections, and retractions per 100 segments.
• Reach and completion: Listener growth and segment completion rates.
• Cost per minute: Production cost versus human-only baselines.
• Trust indicators: Complaint volume, disclosure comprehension, and survey scores.
• Accessibility: Caption accuracy and availability across channels.

Transparency starter kit

Use clear, consistent disclosures across radio, web, and social. Example:

On-air: "This segment is delivered by an AI voice. Editorial content is prepared and verified by our newsroom."

Web/social: "AI voice used for delivery. Human editors are responsible for content and accuracy. Contact [team email] for questions or corrections."

Procurement questions for AI voice vendors

• What languages and dialects are supported? Can we create and manage a custom pronunciation dictionary?
• What logging and audit features exist for scripts, approvals, and changes?
• How are training data and voices sourced and licensed? Are there usage restrictions for government content?
• What controls prevent unauthorized voice cloning or model drift?
• What uptime, latency, and support SLAs are guaranteed for live broadcast?
• How is personal data handled and protected end-to-end?

Policy checklist for agencies

• Publish an AI use policy for broadcast and digital channels.
• Mandate human review and sign-off for all AI-read scripts.
• Maintain a public corrections log and archive of AI-delivered segments.
• Disclose AI use on-air and online, every time.
• Run periodic third-party audits for bias, accuracy, and accessibility.
• Coordinate with HR on role changes, training, and labor considerations.

Bottom line

AI voices can extend your reach and speed, but they do not replace editorial judgment. Treat them as delivery tools. Keep humans accountable for facts, tone, and public trust.

For governance references, see the NIST AI Risk Management Framework here and the Philippines National Privacy Commission resources here.

If your team needs structured upskilling for AI in public communication, explore role-based learning paths at Complete AI Training.