A new UN Women analysis of 133 artificial intelligence systems has found that 44 per cent display gender bias, while more than one in four exhibit both gender and racial bias. The finding lands as separate McKinsey data shows 78 per cent of global organisations now use some form of AI, a figure still climbing with the spread of generative models. Viewed together, the numbers capture a defining tension: the technology is being woven into economic and institutional life at speed, yet the data it is trained on continues to encode decades of unequal representation, routinely associating women with home and caregiving and men with leadership and careers.

That pattern-matching logic, which lacks any human understanding of values or ethics, is the same mechanism enabling a surge of low-quality AI-generated content across social platforms. YouTube, TikTok, Substack and Pinterest have all moved in recent months to label synthetic media or give users controls to limit it, while cybersecurity specialists in Latin America report that generative AI is lowering the barrier to entry for cybercrime, allowing attackers to automate phishing, produce convincing deepfakes and coordinate ransomware campaigns with minimal technical skill.

Regional responses are now hardening. The European Union has enacted its AI Act, and the Association of Southeast Asian Nations is building a regional AI safety network ahead of a Digital Economy Agreement scheduled for signature at the November ASEAN summit. In Brazil, where IDC projects AI-related investment will reach US$3.4 billion this year, logistics platforms are using machine learning to cut delivery costs and marketing firms report that 70 per cent of professionals already deploy AI for content and campaign personalisation. Yet the International Labour Organization estimates that one in four workers globally is in a role highly exposed to generative AI, with administrative, customer-service and translation jobs facing the most immediate transformation. Universities in Ghana are confronting a parallel challenge as students turn to AI for assignments, prompting calls to redesign assessments around critical thinking rather than recall.

Auditors in Australia have been warned that the nature of assurance is changing and that public trust in government AI will depend on competent oversight of automated decision-making. Meanwhile, a consumer counter-movement is gaining visibility: the newsletter platform Substack has integrated an AI-detection tool, and some brands are beginning to treat “made by humans” as a marketable attribute. The next factual milestone to watch is the ASEAN summit in November, where the digital economy pact is expected to establish binding regional principles on transparency, accountability and data governance for artificial intelligence.