新加坡訊：新加坡資訊通信媒體發展局（IMDA）因X和TikTok在識別與清除有害網絡內容方面存在「嚴重缺陷」，正式向兩家平台發出警告信，並將其納入強化監管名單。
IMDA於3月31日發布的《2025年在線安全評估報告》指出，X未能有效遏制涉及新加坡用戶的兒童性剝削與虐待材料（CSEM），而TikTok則在識別和清除恐怖主義相關內容上出現重大漏洞。
根據IMDA定義，警告信是對受監管實體在監管中發現嚴重缺陷的正式譴責，並要求其立即採取補救措施。在強化監管框架下，兩家平台必須定期向IMDA提交整改進展報告，直至監管方確認問題已徹底解決。
為提升檢測能力，兩家平台需引入AI技術，識別不良分子針對新加坡用戶使用的特定關鍵詞與戰術。同時，雙方承諾培訓人工審核團隊，以更精準識別與新加坡相關的CSEM與恐怖主義內容，並加強與科技社群及監管機構的信息共享。
此外，兩家平台須於6月30日前向IMDA提交支持性數據，證明其整改措施的有效性。
IMDA警告：「若X或TikTok無法證明已有效提升對CSEM與恐怖內容的防控能力，我們將毫不猶豫地啟動《廣播法》下的進一步監管行動。」根據該法案，違規平台最高可被處以100萬新元罰款，屢犯者還將面臨持續追罰。
CSEM案例激增，X被動應對成短板
報告顯示，X未能遏制涉及新加坡用戶的CSEM案例激增——2024年僅33起，2025年飆升至73起，增幅超120%。儘管IMDA早在2024年已向其通報問題並提供檢測指標，X仍僅在收到監管方 flagged 後才刪除內容。
其中，68%的案例為協同網絡帳號發布含CSEM關鍵詞的連結，引導用戶跳轉至外部託管網站；其餘32%為用戶自創內容，涉及疑似18歲以下新加坡青少年發布的露骨性圖像，其個人資料中常包含明顯年齡標識（如出生年份）。
更令人擔憂的是，這73起案例平均在平台上滯留9至31周，IMDA直言：「X擁有充足時間主動發現並清除這些內容。」
TikTok恐怖內容「隱身」成謎
2024年，IMDA首次在TikTok上發現新加坡帳號傳播的恐怖主義內容，2025年共發現17起未被主動識別的案例。部分視頻通過剪輯手段，將恐怖組織的影像或音頻嵌入看似無害的內容中，甚至將恐怖音頻偽裝成「原創音效」，使其被納入平台音效庫，供其他用戶二次傳播。
更諷刺的是，即便用戶舉報，TikTok系統仍曾誤判為「合規內容」，直至IMDA介入才被刪除——暴露出其用戶舉報機制的嚴重失效。
兒童安全仍存隱患
儘管Facebook、YouTube、Instagram和TikTok在兒童安全功能上有所進步，但報告仍指出：兒童仍可輕易接觸到裸露、性主題內容。Facebook與YouTube存在兒童帳號瀏覽成人內容的漏洞；HardwareZone的年齡限制在2025年1月底前極易繞過。
IMDA強調：「在日益複雜的網絡風險環境中，社交平台必須持續優先提升兒童安全措施的全面性與有效性，以最大限度減少兒童接觸有害內容的機會。」
平台評分：TikTok下滑，HardwareZone逆襲
六家平台（Facebook、HardwareZone、Instagram、TikTok、X、YouTube）均未獲得滿分5星，但Facebook、Instagram和YouTube因響應用戶報告速度提升而保持穩定。HardwareZone更以「用戶舉報與處理」滿分成為唯一全星平台。
然而，TikTok整體評分因舉報機制退化而下滑；Instagram與TikTok在兒童安全措施上斬獲滿分，而HardwareZone僅得2.5星。
X雖在限制未成年人接觸成人內容上取得進展，但2025年仍未能主動識別CSEM，暴露其系統性短板。
透明度仍是短板
Facebook、Instagram和YouTube未能提供新加坡本地化的用戶報告響應數據，使家長和用戶難以準確評估平台安全性。
IMDA表示，將持續密切監控平台動態，並根據網絡威脅演變，不斷優化監管框架，同時正深入研究如何進一步強化兒童在線安全的強制性要求。
SHARP RISE IN HARMFUL CONTENT
IMDA found that X failed to curb a sharp rise in CSEM cases involving Singapore users. The number of such cases more than doubled from 33 in 2024 to 73 in 2025, despite the regulator informing the platform of the issue in 2024 and sharing detection indicators with them.
The cases of content sharing or linking to CSEM – 68 per cent of the 2025 cases – involved coordinated networks of accounts using terms commonly associated with CSEM, with links that direct Singapore users to external sites hosting CSEM.
The remaining 32 per cent of 2025 cases were self-generated CSEM. These involved accounts sharing posts with self-generated explicit sexual imagery from users in Singapore who were purportedly under 18.
These users included commonly known number patterns on their profile indicating their date of birth or age.
The report also found that a majority of these 73 cases were also on the platform for an extended period of time, ranging from nine to 31 weeks on average.
"X therefore had ample time to proactively detect and remove these cases," said IMDA in its report.
IMDA pointed out that all 73 cases violated X’s own policies against CSEM however, X only removed them after they were flagged by IMDA.
Last year, IMDA detected terrorism content shared by Singapore-based accounts for the first time.
There were 17 such cases of terrorism content shared by Singapore-based accounts on TikTok that were not proactively detected and removed in 2025.
Some videos used edited footage or audio linked to known terrorist organisations that were blended with benign content. In some cases, the terrorism-related audio was also concealed under the platform's "original sound" label, which adds it to TikTok's database for others to use in their posts as well.
According to the report, there were instances where TikTok initially assessed the content as non-violating even after user reports. They were removed only after IMDA intervened.
This demonstrates the weakness of TikTok's user reporting system, said the authority.
IMDA stressed that CSEM and terrorism content are 「very egregious harms」 that must be proactively detected and swiftly removed before users encounter them, as required under Singapore’s Code of Practice for Online Safety for social media services.
In response to IMDA’s findings, TikTok said it does not allow violent and hateful organisations or individuals on its platform.
It added that in the fourth quarter of 2025, its proactive systems successfully removed over 99 per cent of content violating its policy before it was reported, with over 93 per cent of content being removed within 24 hours.
「As violent extremist methodologies and evasion techniques continue to evolve, so do we. We continually evaluate and strengthen our policies and systems, and are committed to partnering with IMDA under the enhanced supervision scheme,」 said TikTok in IMDA's report.
X said it maintains a zero tolerance policy towards child sexual exploitation material, adding that it is deeply committed to child safety.
"We have made meaningful progress in restricting minors』 access to adult content on X, and warmly welcome IMDA’s recognition of this," said the platform in IMDA's report.
