Algorithms alone cannot shield children from online harms, says expert

 Cybersecurity practitioner Murugason R Thangaratnam says coded language, evolving tactics, and subtle grooming behaviour can still bypass detection.

The Online Safety Act 2025 now requires social media platform users to upload official government-issued identity documents for age verification purposes. (Envato Elements pic)

PETALING JAYA:

Algorithm moderation alone cannot fully protect children from online harms, a cybersecurity expert has warned.

Cybersecurity practitioner Murugason R Thangaratnam said safer algorithms can reduce exposure to harmful content but warned that they were not a complete remedy.

Murugason R Thangaratnam.

“Even with the most advanced moderation systems, it is unrealistically optimistic to assume that minors will be entirely shielded from harmful communities, grooming attempts or dangerous content online,” he told FMT.

Algorithmic moderation involves using automated systems to filter or remove harmful content on social media platforms.

Murugason said heavily moderated algorithms could significantly reduce young users’ exposure to graphic violence, hate speech and exploitative material by creating a “default environment” less likely to normalise harmful behaviour.

However, he said these algorithms could still be evaded through coded language and evolving tactics, while some risks—including subtle grooming behaviour and emotionally manipulative content—may go undetected if they fall below set thresholds.

Murugason said social media harms extend beyond harmful content itself, pointing to addictive platform design and engagement-driven systems.

He said features designed to maximise attention and repeated usage remain a concern, particularly for younger users.

“A truly ‘child-safe’ algorithm is both technically and ethically problematic. Algorithms learn from historical data that encodes real-world biases and harms, and the internet’s openness and diversity ensure new risks will keep arising,” he said.

Murugason said that defining what constitutes child‑safe content may be culturally contested. He also warned that excessive moderation could risk suppressing legitimate expression.

Effective June 1, users of social media platforms were required to upload official government-issued documents for age verification purposes following the implementation of two new codes under the Online Safety Act 2025.

The two codes introduced by the Malaysian Communications and Multimedia Commission — the Child Protection Code (CPC) and the Risk Mitigation Code — are aimed at preventing individuals under 16 from opening social media accounts.

While describing the CPC as promising, Murugason said its effectiveness would ultimately depend on strong enforcement, clear standards and continuous oversight.

Nevertheless, he described the under-16 social media ban as potentially more effective than relying solely on algorithm-focused regulation. He said it created a “clear, enforceable boundary” that removed children from “the most complex and harmful corners of social media”.

“It sends a powerful message that children are not fair game for commercial exploitation or algorithmic manipulation,” he said.

Dr Amer Siddiq Amer Nordin.

Universiti Malaya psychologist Dr Amer Siddiq Amer Nordin said children and younger teenagers are generally not developmentally capable of resisting platforms designed to encourage prolonged use.

“If they are unable to give informed consent or fully understand consequences, then they are also unlikely to resist algorithms specifically designed to maximise screen time and nudge users towards endless scrolling,” he said.

Amer, a consultant psychiatrist, noted that changing user behaviour was often difficult, especially among children already heavily reliant on social media, making age-based restrictions more necessary.

Both experts agreed that stricter algorithm regulation would ultimately complement the ban and aid in protecting children from exposure to harmful digital environments. - FMT