The artificial intelligence industry has pivoted sharply from raw capability to cost efficiency. Within a single week, OpenAI, Meta and SpaceXAI each launched new models whose primary selling point was not superior reasoning but lower token consumption. OpenAI’s GPT-5.6 is engineered to perform more work with fewer tokens; SpaceXAI’s Grok 4.5 claims double the token efficiency of comparable models; and Meta’s Mark Zuckerberg is offering Muse Spark 1.1 at what he calls a fraction of competitors’ prices. The shift follows months of corporate sticker shock, as enterprises that had encouraged maximal AI use began receiving bills running into millions of dollars and imposed tighter controls.

The mechanism driving this price war is itself becoming a flashpoint. Competitors are distilling intelligence from top-tier models—harvesting outputs at scale to train cheaper alternatives—in what Anthropic, OpenAI and Google describe as a cybersecurity threat. Yet the practice mirrors the very web-scraping that AI firms have long defended as fair use. Efforts to lock down access have triggered an escalating cat-and-mouse game, with researchers noting that once information is online, determined actors will find ways to extract it. The dispute exposes a structural tension: the same open architecture that enabled rapid AI development now makes enforcing terms of service extraordinarily difficult.

That enforcement gap is not confined to technology. In Africa, anti-corruption organisations marking African Anti-Corruption Day warned that the continent possesses ample laws and institutions but loses an estimated $88.6 billion annually to illicit financial flows because powerful individuals escape accountability. In Iran, banking disruptions have become so routine that the Chamber of Guilds publicly criticised the central bank for treating systemic outages as minor technical glitches rather than governance failures. Kenya’s transport regulator was forced to suspend mandatory vehicle inspections after a court challenge and a Senate committee found it had bypassed legislative scrutiny and could not even state how many vehicles were on the road. In Nigeria, the bar association’s election is proceeding amid legal disputes over digital voter identification, with the electoral chairman refusing to resign despite calls from senior lawyers.

Viewed from Abidjan, where the African Economic Conference opened with calls for the continent to move from dependency to agency, the pattern is clear: institutions are struggling to match the speed of technological and regulatory change. The conference’s focus on strengthening Africa’s geopolitical resilience and trade architecture reflects a broader recognition that capacity must catch up with ambition. The next factual milestones will be the Nigerian Bar Association election on 18 July, the Kenyan High Court’s substantive ruling on the inspection regulations, and the AI industry’s response to distillation litigation—each a test of whether enforcement can finally close the gap.