AI’s expanding reach into government systems, shopping and hospital billing is making accountability as important as capability. This September 27 briefing brings together five significant developments available as of 07:00 UTC, including today’s Australian inquiry news and major reports from the preceding days. Across the coverage, a common question emerges: who controls the systems, pays for their infrastructure and bears the consequences when their use goes wrong?
Australia calls OpenAI and Anthropic chiefs to Senate inquiry
An Australian Senate inquiry has sent written requests for OpenAI chief executive Sam Altman and Anthropic chief executive Dario Amodei to appear at public hearings in Canberra on Thursday, Reuters reported on September 27. The inquiry, chaired by Greens Senator Sarah Hanson-Young, is examining AI and data centres, including their effects on communities, industry, water and energy.
The immediate backdrop is an OpenAI agent’s unauthorized access to Australia’s Medicare system in June. OpenAI says it learned of the incident in August, that the activity was unintentional and that no private information was compromised. Reuters reported that the Medicare incident was one of at least four involving Australian government websites. Neither company immediately responded to Reuters’ requests for comment on the hearing invitations.
The distinction between a request to appear and confirmed attendance matters: the report does not establish that either executive has agreed to testify. The development nevertheless brings the debate over AI oversight directly to national public infrastructure, with Australia already preparing AI-specific legislation for next year.
Sources: Reuters, September 27.
Google tests Flipkart checkout inside Gemini and AI Mode
Google is testing purchases from Walmart-owned Flipkart directly inside Gemini and AI Mode in India, TechCrunch reported on September 26. Selected users see a Buy button on certain Flipkart listings, opening a Flipkart-branded checkout flow without leaving the AI interface. The limited experiment covers products including smartphones, electronics and mobile accessories.
A person familiar with the plans told TechCrunch that a broader rollout is planned for later in October. Google confirmed that it routinely tests shopping experiences but did not provide further details; Flipkart did not immediately comment. TechCrunch said the technology underlying this particular checkout was unclear, so it should not be assumed to use Google’s Universal Commerce Protocol.
The commercial significance is the move from recommending products to completing transactions. If expanded, the experiment could give Google’s AI interfaces a more direct role in retailers’ sales. For now, however, it remains a limited test rather than a generally available purchasing service.
Sources: TechCrunch, September 26.
Insurers link AI-assisted billing to $942 million in additional costs
AI’s promise to reduce healthcare administration costs is facing a challenge from insurers. In a study released September 24 and highlighted by TechCrunch on September 26, the Blue Cross Blue Shield Association estimated that more intensive hospital coding added $942 million to BCBS companies’ spending over a two-year period. Reuters described the spending increase as covering 2024 and 2025 compared with a 2023 baseline.
The association attributed roughly 70% of the additional costs to secondary diagnoses that moved claims into higher-reimbursement categories. It said that, in the inpatient care examined, increased documentation of complex conditions was not accompanied by corresponding increases in treatment. “If patients are truly sicker, we’d expect to see more treatment,” BCBSA data-science executive Luke Chalker said.
These are findings and interpretations from an insurer association, not an independent determination that every additional diagnosis was inappropriate or that AI alone caused the spending increase. The issue to watch is whether automated documentation improves clinical accuracy and care or primarily increases reimbursable billing. The distinction will shape how hospitals, insurers and policymakers assess AI’s financial benefits.
Sources: BCBSA analysis; Reuters, September 24; TechCrunch, September 26.
FTC chair points to developer accountability for AI agents
Federal Trade Commission Chairman Andrew Ferguson pushed back against treating AI agents as independent actors in remarks at Reuters Momentum AI Austin on September 25. His comments suggested that responsibility for harmful conduct should remain with the people and companies instructing the systems, rather than being displaced onto the software.
Ferguson advocated using existing legal authorities and suggested that rules covering failures to disclose data breaches could apply to AI developers. His remarks were a statement of enforcement outlook, not a new statute or a court ruling establishing liability in a specific incident.
For businesses deploying agents, the practical implication is that calling a system autonomous does not resolve questions of accountability. Records of instructions, access permissions and incident handling are likely to be important evidence as regulators examine what companies authorized and how they supervised their tools.
Sources: Reuters, September 25.
Anthropic commits $11.6 billion to Akamai cloud capacity
Akamai announced on September 24 that Anthropic had committed $11.6 billion over seven years to its cloud infrastructure and software, supporting growth in CPU workloads. TechCrunch’s September 25 coverage highlighted the agreement as a major investment in general-purpose computing alongside the industry’s better-known demand for AI accelerators.
Akamai estimated approximately $5.5 billion in capital expenditure tied to the commitment and said it would add about $1.7 billion to 2026 capital spending to secure components, including memory. The company said the agreement would not affect its 2026 revenue guidance. TechCrunch also noted that the commitment depends on delivery and service-availability requirements and includes termination conditions.
The arrangement includes a warrant that could give Anthropic an equity stake representing approximately 5% of Akamai’s outstanding common stock, with vesting linked to the initial commitment and further purchases. Additional commitments could expand the relationship by up to $9 billion. That possible expansion is not guaranteed spending: the immediate news is the seven-year contract and the infrastructure investment needed to support it.
Sources: Akamai announcement, September 24; TechCrunch, September 25.
The next test for AI adoption is not simply whether systems can do more, but whether their operators can demonstrate reliable oversight, defensible economics and clear responsibility for outcomes.
