AI Melawan Disinformasi, Fitnah, dan Kebencian

Poin-Poin Utama

• • Mafindo recorded 1,593 hoaxes between October 2024 and October 2025, mostly AI-generated.
• • Wamenkomdigi Nezar Patria cited Sensity AI data in September 2025 showing deepfake content rose 550% in five years.
• • Nine out of ten deepfakes are used for harmful purposes.
• • A Yogyakarta cybersecurity workshop in early 2026 reported a 1,400% surge in deepfake attacks within one year.
• • AI-based phishing attacks achieved success rates between 54% and 60%.
• • Mafindo chair Septiaji Eko Nugroho stated deepfakes in Indonesia are mainly used for fraud and political opinion manipulation.
• • Bawaslu chair Rahmat Bagja warned AI enables near-indistinguishable content production ahead of the 2029 election.
• • At least 11% of young women aged 15-29 have experienced online gender-based violence, mostly via deepfake.
• • Mafindo's 2025 study showed transformer-based language models classify Indonesian hoaxes with over 90% accuracy.
• • Jurnal Khatulistiwa Informatika (June 2026) reported an autonomous AI hoax detection system achieving 89.3% accuracy.
• • The autonomous detection system can show evidence behind its decisions.
• • The article frames AI as both beneficial and harmful simultaneously.
• • Social media algorithms prioritize sensational hoax content because it is deemed engaging.
• • Coordinated bot networks and micro-targeting are used to influence specific voter groups.
• • The article calls for using ethical AI (AI Baik) to combat malicious AI (AI Jahat).