By Halim Mohammad (Dhaka Bureau)

DHAKA — The exponential growth of artificial intelligence (AI) users in Bangladesh has brought with it an alarming surge in tech-driven crimes. While AI platforms are streamlining daily tasks, the malicious use of these technologies is fueling widespread misinformation, rumors, and severe harassment of ordinary citizens. In response, experts are urgently calling for the implementation of a comprehensive national and global “AI Policy” to curb institutional misuse and mitigate these negative impacts.

A recent survey of nearly 100 regular AI users in Dhaka highlighted both the popularity and the pitfalls of the technology. According to the respondents, at least eight major text and content-writing tools are heavily utilized in Bangladesh, including ChatGPT, Claude AI, Google Gemini, Copy.ai, and Grok AI.

Visual AI generation has also captivated the country’s youth. Roughly a dozen advanced image and video tools—such as Copilot, Midjourney, DALL-E 3, Sora 2, Runway Gen 4.5, and Google Veo 3.0—are seeing increased adoption, alongside standard conversational chatbots. However, high premium subscription costs keep the number of paid users relatively low. Interestingly, several surveyed students admitted they frequently interact with AI simply as a way to cope with daily loneliness.

Yet, the technology’s dark side remains a prominent issue. Highlighting the flaws of unregulated use, one student noted: “AI often just gives a generalized concept, but it frequently generates hallucinated or incorrect information. When we use it without verifying, we end up in deep trouble.”

The creation of seamless, AI-generated videos—often in the form of deepfakes—has made it increasingly difficult for the public to separate truth from fiction. This intense confusion across social media is triggering volatile, real-world incidents, threatening the nation’s political landscape, education system, job market, and cinema industry.

The local fact-checking organization Rumor Scanner underscored this threat with harrowing statistics for 2025: 68 severe rumors and misinformation campaigns targeted 29 prominent women in the Bangladeshi showbiz industry. Alarmingly, 50 of these 68 smear campaigns utilized advanced AI video technology.

When asked to identify the premier AI tool on the market, Professor Khandakar Mamun, Vice President of AI at the Washington University of Science and Technology, explained: “The platform we consider the best today might be completely overshadowed by an even more advanced medium hitting the market just a month from now. Therefore, it is technologically impossible to single out one tool as the absolute best. Its utility depends entirely on the nature of the task.”

Echoing this pragmatic view, AI trainer Obaidur Rahman Robin added: “An individual will ultimately choose and utilize whichever medium best aligns with their specific needs and personal convenience.”

To combat this rising tide of digital crime, tech analysts stress the need for strict AI regulations—with a major caveat. They strongly caution that authorities must not weaponize these laws to suppress dissent. To prevent a “rumor prevention law” from deteriorating into a “criticism-suppression law,” legislation must clearly define and distinguish between legitimate criticism, political sarcasm, satire, hate speech, misinformation, and deliberate disinformation.

Experts point out that arrests and criminal prosecution should be a strict last resort. Instead, the primary strategies should include:
1. Promptly issuing counter-narratives.
2. Independent and rapid fact-checking.
3. Ensuring the government’s explanation and the citizen’s right to reply.
4. Enabling swift reporting mechanisms on digital platforms alongside civil remedies.

Accountability must also extend to social media conglomerates like Facebook, TikTok, and X. Analysts argue these platforms must bolster their transparency and local language moderation to effectively combat fake photocards, bot networks, unlabelled AI videos, and communal content. Referencing Meta’s past moderation failures that facilitated ethnic violence in Myanmar, experts warned that delayed platform interventions can lead to actual real-world casualties.

Furthermore, rather than the state acting as the sole arbiter of truth, there must be a concerted push to empower independent research, media collaborations, fact-checking initiatives, and civic media literacy.

“As long as the proper and institutional utilization of AI is not firmly established in our country, we will not be able to bring its malicious aspects under control,” Professor Mamun concluded. Ultimately, as Robin emphasized, bringing AI usage under a forward-looking, well-defined legal framework is now absolutely mandatory to safeguard Bangladesh’s broader cyber security.

*Halim Mohammad is a Senior Journalist and Contributor for the Pressenza Dhaka Bureau.*