US export controls on Anthropic’s Claude Fable 5 and Mythos 5 lifted
Plus, China's plan to save jobs from AI
Welcome to the latest edition of ASPI’s Cyber & Tech Digest.
Each week, ASPI curates and contextualises the most important developments in cyber, technology, and geopolitics — highlighting what matters and why.
This edition covers the period: 27 June 2026 to 3 July 2026.
Follow the Australian Strategic Policy Institute on Bluesky, LinkedIn, and X.
What We’re Tracking
US export controls on Anthropic’s Claude Fable 5 and Mythos 5 lifted
What happened: On 30 June, the US Commerce Department lifted export controls on Anthropic’s Claude Fable 5 and Mythos 5, less than three weeks after ordering Anthropic to suspend access to the models on national security grounds. Anthropic said it had received notice that the controls had been lifted and would begin restoring access from 1 July. Commerce Secretary Howard Lutnick’s letter withdrew the export controls but said the department could re-evaluate the decision if circumstances change or Anthropic fails to meet its commitments.
The reversal follows a period of intense negotiation between Anthropic and US officials after the 12 June directive forced the company to disable Fable 5 and Mythos 5 globally, because it could not reliably restrict access to US nationals. Anthropic had argued at the time that the government had not provided detailed evidence of its concern and that the alleged jailbreak risk was narrow, non-universal and comparable to risks in other public models.
The deal appears to rest on Anthropic agreeing to detect and address security risks, work with the US government on release protocols for current and future models, and report malicious activity. Anthropic has also reportedly implemented a new safeguard that blocks the jailbreak technique officials were worried about in 99% of cases.
The immediate context is broader than Anthropic. The Trump administration’s 2 June AI executive order created a voluntary framework for developers to provide the US government with access to ‘covered frontier models’ for up to 30 days before release to trusted partners co-selected by both the developers and the government.
Why we’re tracking this: The reversal matters because it shows that frontier-AI restrictions can be imposed abruptly, but also negotiated away through safeguards, monitoring and government-access arrangements. That makes the episode less a one-off ban than an early test of a new governance model: conditional access to frontier AI. For allies such as Australia, the reliability concern remains. Even if Fable and Mythos are restored, access now looks more contingent on US national-security judgements, security protocols and trusted-partner frameworks. The issue is therefore not only whether models are available, but who gets assured access when the most capable systems become strategically important.
What people are saying:
‘The lesson of the past two weeks is not that the United States should stop competing. It is that the two reflexes now defining U.S. strategy - walling off the world through export controls on deployed AI models and refusing to govern at home with reasonable regulation - has placed the U.S. AI economy on a very shaky foundation.’ — Alvin Wang Graylin and Jon J. Rosenwasser, Lawfare
‘The administration needs to answer three questions: Who decides when the threat is serious enough? On what quality of evidence? And through what process? These are as much technical questions as constitutional ones.’ — Elly Rostoum, Senior Fellow at the Center for European Policy Analysis
‘This was a huge own goal for the US, and we will see how bad US models get over the next six months and if Chinese models become noticeably better for cyber work.’ — Alex Stamos, Chief Security Officer at Corridor
My view: The lifting of export controls does not undo the strategic lesson of the original directive. The US government appears to be moving from a blunt emergency restriction toward a managed-access model for frontier AI: stronger safeguards, closer government visibility, pre-release review, trusted-partner access, and the possibility that controls can return if companies fail to satisfy national security expectations.
That is probably more sustainable than an open-ended ban. But it also confirms that frontier models are no longer being treated as ordinary commercial software. They are increasingly viewed as dual-use capabilities whose distribution may depend on government confidence in the provider, the model’s safeguards and the users who receive access.
For US partners, the reassurance remains limited. Access to Fable and Mythos may be restored, but the precedent remains that the US can intervene directly in model availability. The next contest will be over rules, not only access: what counts as a dangerous cyber capability, who defines a trusted partner, what evidence is required before restrictions are imposed, and whether allies have any say in those decisions. Australia should therefore pursue privileged access arrangements with the US, but also invest in domestic AI assurance, sovereign evaluation capability and diversified partnerships so that frontier-AI dependence does not become a single point of strategic failure.
— Dr Gatra Priyandita, CTS
China’s plan to save jobs from AI
What happened: China’s tech workers have long dreaded the word youhua 优化, ‘optimisation,’ corporate shorthand for lay-offs. This year it carries a new undertone, SCMP reports. Keeping your job used to depend on doing it well. Now it depends on whether AI can do it too. After viral (and denied) rumours that Meituan, the food-delivery giant, would cut up to half its product roles, a quieter retrenchment appears to be under way across Baidu, Xiaomi, ByteDance and Alibaba. Senior Tencent executive Dowson Tong said most of the company’s code this year was AI-generated. Meanwhile, the Financial Times reports AI is pushing China’s factory robots beyond autos into garments and footwear. Courts are beginning to draw the line. In April, a Hangzhou court ruled a fintech firm unlawfully sacked a quality-control supervisor it replaced with AI, ordering more than 260,000 yuan in compensation.
Why we’re tracking this: China is running the world’s largest experiment in state-managed AI labour displacement: maximum-speed adoption, paired with shock absorbers only a party-state can deploy. Whether that combination holds has consequences for China’s economy, its stability, and every government now drafting AI employment policy.
What people are saying:
‘The development of artificial intelligence technology should be applied to liberating labor, promoting employment and improving people’s livelihood.’ — Hangzhou Intermediate People’s Court
‘We must do everything possible to protect the livelihoods of hundreds of thousands of employees, including the jobs of our blue-collar brothers.’ — Richard Liu, founder and chairman of JD.com
‘To say that companies are laying people off just because AI has improved workflows is basically a lie.’ — Tuxi, veteran AI developer at several major Chinese tech firms
My view: Until recently, the party line on AI-driven automation focused almost exclusively on the upside. The ‘AI Plus’ plan targets AI integration across more than 70 per cent of major industries within five years, and last year China installed more industrial robots than the rest of the world combined. But the party fears educated, unemployed urban youth more than any other constituency, and so the rhetoric has shifted. Beijing has stopped pretending that a labour shock isn’t coming. Courts are elevating rulings against AI-justified sackings into ‘representative cases,’ the new five-year plan commits to ‘comprehensively address’ AI’s impact on employment, and JD.com is promising to retrain 700,000 blue-collar workers rather than shed them. Whether any of this actually cushions the shock is another matter. The pledges remain untested, the rulings are case-by-case, and the one tool Beijing refuses to reach for is welfare. Xi Jinping, the leader of a nominally communist party, has repeatedly warned against falling into the trap of a ‘welfarism’ that breeds ‘lazy people.’ Displaced workers will be retrained and redirected, but they will not be paid not to work.
Meanwhile, the Bank for International Settlements, which advises the world’s central banks, has warned that the AI boom now ranks alongside railways and dotcom: technological breakthroughs that attracted ‘capital in excess of what commercial returns could ultimately justify,’ and ended in recessions. The US appears to be stuck in a trap — either AI successfully wipes out enough jobs to justify the US$1 trillion the hyperscalers are pouring into it, or it all goes belly up. Either way, a big shock is on its way. Beijing is at least making the right noises about preparing for it.
— Fergus Ryan, CTS
What We’re Watching
A weekly scan of notable developments we’re tracking across technology, policy, and geopolitics.
⸻
🚀 Strategic competition
Job postings suggest DeepSeek is developing an agentic model capable of finding software vulnerabilities and building attack paths, according to Alex Colville as published in The Strategist.
China’s AI sector continues to narrow the gap with US frontier labs. Zhipu AI’s open-weight model GLM-5.2 can match Anthropic’s Mythos on some vulnerability-finding tasks, according to The Wall Street Journal, while 360 Security Technology has launched a comparable bug-finding tool called Tulongfeng.
DeepSeek has released DSpark, an upgrade to its V4 model designed to speed up inference and cut computing costs using a lightweight draft model and semi-autoregressive generation. The company has open-sourced the technology on GitHub and HuggingFace.
Gerald Mako provides an analysis describing how China’s Joint Logistic Support Force is integrating AI into military logistics under a military-civil fusion model that treats commercial transport and digital infrastructure as extensions of the military supply network.
Austria has urged the European Union to explore hosting Anthropic within the bloc to avoid being cut off from advanced AI models following US access restrictions. State Secretary for Digitalisation Alexander Proell made the proposal in a letter to EU Technology Commissioner Henna Virkkunen. The move comes as the European Commission advances plans to build up domestic AI, cloud and semiconductor capacity.
Tokyo-based Sakana AI has released Fugu, an orchestration model aimed at businesses and governments seeking to reduce exposure to US export controls.
In The New York Times, a report says China is focused not only on matching US model capability but also on managing the domestic labour disruption AI adoption is causing.
Despite tightening geofencing, Wired reports that users and companies in China continue to circumvent Anthropic’s geolocation restrictions on Claude through proxy services and fake identities.
China’s compute build-out continues on both the training and hardware-access fronts. The National Supercomputing Center in Shenzhen has topped the June Top500 rankings with its ‘LineShine’ system, built on domestically developed LingKun LX2 processors.
Research firm WireScreen found that Tsinghua University acquired advanced Nvidia accelerators, including A100, H100 and RTX 5090 chips, between November 2025 and May 2026 despite successive rounds of US export controls.
Taiwanese authorities have raided the offices of Super Micro Computer and several local affiliates as part of a probe into the alleged smuggling of Nvidia chips into China via the company’s servers. Prosecutors have since detained two Super Micro employees and placed two others on bail, with distributor Albatron Technology and data-centre operator Chief Telecom also implicated in what The Japan Times describes as Taiwan’s first public crackdown on AI chip diversion.
Japan’s Ministry of Internal Affairs and Communications will launch eight digital infrastructure projects in Taiwan and India under a new ‘digital corridor’ linked to Prime Minister Takaichi Sanae’s updated Free and Open Indo-Pacific strategy. The projects, covering undersea cables, satellite communications and next-generation infrastructure, form part of Nikkei Asia’s reporting on growing Indo-Pacific digital economic security cooperation.
OpenAI chief executive Sam Altman has proposed a US-led international framework to govern AI, arguing in a Financial Times op-ed that independent global standards are needed to assess AI capabilities and risks. He compared the proposed mechanism to international bodies overseeing aviation safety and nuclear material.
Amazon says it has deployed 396 low-Earth-orbit satellites for its Project Leo broadband network, enough to begin continuous service across initial coverage areas, though it remains behind SpaceX’s Starlink rollout pace.
⸻
🧠 AI models, agents & compute
OpenAI’s GPT-5.6 system card describes GPT-5.6 Sol as a substantial improvement over GPT-5.5 but below Anthropic’s Mythos on frontier capabilities, according to The Zvi, which flagged improved cybersecurity and biological capabilities alongside new reasoning modes and concerns about agentic misalignment and evaluation gaming.
Microsoft has launched the Microsoft Frontier Company, a US$2.5 billion initiative that will embed more than 6,000 engineers inside customer organisations to build and operate AI systems.
SoftBank Group and its telecom unit will establish SB Neo Inc. to provide AI chips and cloud computing services to US companies, aiming for around 10GW of capacity.
Google has reportedly restricted Meta’s access to Gemini computing capacity after being unable to meet demand, according to a Reuters report citing the Financial Times, delaying some of Meta’s internal AI projects. Advanced chip packaging has meanwhile become a critical bottleneck for AI computing, with Taiwan Semiconductor Manufacturing Company accounting for nearly all advanced packaging used in leading AI chips, The New York Times reports, after the cancellation of a proposed US$1.1 billion packaging research centre.
A global DRAM shortage driven by AI demand is sharply increasing costs across consumer electronics, CNBC reports, with Apple and Microsoft announcing price rises linked to the shortage. Analyst Ming-Chi Kuo said the imbalance will keep widening through 2027 as capacity shifts from consumer devices to AI data centres, and said Apple is lobbying Washington to keep Chinese memory maker CXMT off the Entity List to secure additional DRAM supply. Separately, The Financial Times reports Apple is seeking approval to buy memory chips from CXMT despite the firm’s presence on the Pentagon’s Chinese Military Company blacklist.
Australian data centre company Firmus plans a major expansion into Indonesia, where it will become a tenant in AI factories being built by DayOne and install up to 170,000 Nvidia chips from late 2026, while continuing planned AI factories in Tasmania, Victoria, NSW, WA and the ACT. Elsewhere, ByteDance is advancing a data centre project in Brazil’s Ceará state estimated to cost US$39 billion that could become its largest complex outside China.
Anthropic is reportedly discussing a custom AI chip with Samsung as it seeks to diversify its computing infrastructure and reduce dependence on Nvidia.
In One Useful Thing, Ethan Mollick argues that frontier AI capability is accelerating rapidly and that AI use is shifting from chatbot-based collaboration toward autonomous agents performing extended tasks with limited human oversight.
⸻
🛡 Cyber posture
Ukraine’s Security Service, working with the FBI, said it uncovered a long-running Russian campaign targeting messaging accounts of officials, military personnel and activists in Ukraine, Europe and the US, using social engineering to impersonate platform support staff. The US State Department has separately offered up to US$10 million for information on Russia-linked groups UNC5792 and UNC4221, which US authorities say are tied to the FSB and have targeted Signal and WhatsApp accounts of officials and journalists.
The US Federal Communications Commission has voted to strengthen oversight of undersea internet cables, banning equipment from China and other foreign adversaries and introducing licensing requirements for terminal equipment operators.
A 19-year-old dual US-Estonian citizen was extradited from Finland to face US charges over cyber intrusions and fraud linked to the Scattered Spider group, including a 2025 breach of a luxury jewellery retailer. The FBI says Scattered Spider has been linked to more than 100 network intrusions and over US$100 million in ransom payments.
Researchers from Searchlight Cyber identified vulnerabilities in WiseTech Global’s CargoWise WebTracker module that could have let attackers impersonate customers and access sensitive logistics and financial data. WiseTech said it had patched the flaw, found no evidence of exploitation, and has retired WebTracker in favour of its newer CargoWise Neo platform.
A developer has accused Anthropic of embedding hidden detection mechanisms in Claude Code that allegedly identify users connecting through China-linked proxies via invisible Unicode characters and date-format changes. As of the end of June, Anthropic had neither confirmed nor denied the claims nor issued a patch.
A security researcher disclosed a vulnerability in Apple’s ‘Hide My Email’ feature that can reveal users’ real email addresses despite the service being designed to conceal them; 404 Media verified the flaw remained exploitable as of the end of June and Apple has not fixed it in over a year.
⸻
🕵️ Surveillance states
Flock Safety’s AI-enabled automated licence plate readers have expanded rapidly across the United States, with more than 100,000 installed nationwide, Engadget reports. The cameras can search for vehicles and people through natural-language queries and have been misused by some police officers to track private individuals, with several cases of errors leading to innocent people being stopped or investigated.
The US Supreme Court ruled 6-3 that police access to cellphone location history held by technology companies constitutes a Fourth Amendment search generally requiring a warrant. The case arose from the prosecution of Virginia bank robbery suspect Okello Chatrie, whose Google location history was obtained via a geofence warrant, and the Court has sent the case back to the Fourth Circuit to assess whether the specific warrant was sufficiently narrow.
In Natto Thoughts, Eugenio Benincasa analyses how China uses cyber operations and private contractors to support transnational repression against critics abroad, citing an alleged China-linked breach of Italy’s Interior Ministry network.
⸻
🧒 Online harms & child safety
Meta instructed contractors working through Covalen to create under-18 personas and send thousands of high-risk prompts involving suicide, sex and eating disorders to rival chatbots including ChatGPT, Gemini and Character.ai, according to a WIRED report cited by LiveMint. A testing round completed in August 2025 reportedly involved more than 45,000 prompts; Meta said the activity was standard safety benchmarking not used to train its own models.
The US House of Representatives passed the Kids Internet and Digital Safety (KIDS) Act with bipartisan support, introducing AI chatbot disclosure requirements, age verification for pornography access and new rules for data brokers handling children’s information. The bill faces opposition in the Senate, where lawmakers back a competing Kids Online Safety Act with a stronger duty-of-care requirement.
TikTok and YouTube have reached settlements with a Florida teenager who alleged social media platforms harmed his mental health through addictive design features, leaving Meta and Snap to face trial in Los Angeles County over consolidated litigation.
The first report by a UN independent scientific panel on AI says the technology offers major potential benefits but also significant risks to mental health, information integrity and social systems, warning that capabilities are outpacing governments’ ability to govern them. The report also flags risks from AI-generated child sexual abuse material and large-scale persuasive content.
⸻
🏛️ Government, procurement & public sector tech
California Governor Gavin Newsom’s administration has signed a deal with Anthropic to expand use of its AI models across state government services, including the state’s digital assistant, DMV customer service and cybersecurity work, despite the Pentagon designating Anthropic a supply-chain risk in March.
New court documents in Anthropic’s lawsuit against the Pentagon reveal tensions between the company and the Trump administration over military use of AI, detailing exchanges between chief executive Dario Amodei and Undersecretary of Defense Emil Michael over deployment guardrails. The dispute was reportedly resolved on 1 July, though The Wall Street Journal notes the underlying policy disagreements remain.
The US Office of Personnel Management and Department of Defense have launched ‘War Force,’ a recruitment initiative seeking software engineers and AI specialists to support the Pentagon’s AI Acceleration Strategy.
CIA Director John Ratcliffe separately said the agency has undertaken a ‘fundamental reshaping’ of its technology approach, elevating cyber functions and cutting acquisition timelines from nearly three years to about six months.
White House budget director Russell Vought has assumed direct oversight of classified US intelligence spending plans following the departure of Amaryllis Fox Kennedy, giving him scrutiny over nearly US$132 billion in proposed spending as the administration seeks to shrink the Office of the Director of National Intelligence.
A new US executive order directs the National Institute of Standards and Technology to develop a classified benchmarking process for frontier AI models within 60 days and requires up to 30 days of government access to certain models before release to trusted partners.
The Trump administration has discussed with SpaceX the possibility of donating company stock to the children’s savings vehicles known as Trump Accounts, which are expected to launch shortly and had already enrolled more than 6 million children. OpenAI is separately reportedly in early-stage talks to give a 5% equity stake to the US government, an idea chief executive Sam Altman has floated extending to other major AI firms.
⸻
🎬 IP, media & creative industries
The US Department of Justice seized nearly 400 domains used to illegally stream FIFA World Cup matches as part of Operation Offsides, targeting infrastructure across Peru, Bulgaria, Croatia, Romania, Poland and Colombia with support from FIFA and anti-piracy groups.
In Spain, aggressive anti-piracy blocking measures aimed at football streaming are causing collateral disruption to legitimate websites, with one Albacete retailer saying his online store is frequently blocked during LaLiga matches despite no connection to pirated content.
Generative AI and ‘vibe-coding’ have fuelled a sharp increase in game production, with 181,000 games released globally in the six months to May, up 118% on iOS year on year, though the largest publishers continue to dominate revenue and downloads.
TIDAL, the Norwegian-American music streaming service, has introduced a policy that will automatically label wholly AI-generated music and block it from earning royalties or direct-to-fan sales, while removing AI tracks that impersonate artists or are linked to fraud.
⸻
🧑⚖️ Courts, enforcement & regulation
The Australian Competition and Consumer Commission has filed proceedings in the Federal Court against Amazon Australia, alleging it used unfair contract terms to unilaterally introduce advertising to Prime Video and charge subscribers an extra A$2.99 a month to stay ad-free. The regulator alleges the terms affected more than one million subscribers and is seeking penalties, consumer redress and other orders.
Google has lost its final appeal against the EU’s €4.1 billion Android antitrust fine, with the Court of Justice of the European Union ruling that the European Commission was justified in finding Google abused Android’s market dominance by imposing restrictions on device makers. The original 2018 fine alleged Google required manufacturers to pre-install Search, Chrome and Play Store while discouraging rival operating systems, and the ruling is expected to influence future abuse-of-dominance cases alongside the bloc’s Digital Markets Act.
Britain’s Competition and Markets Authority has proposed allowing app developers to direct users to alternative payment options outside Apple and Google’s app stores, and is considering requiring Apple to open its NFC technology to rival payment services.
Alibaba and its US payment processor, AUS Merchant Services, agreed to pay US$600 million and entered non-prosecution agreements with the US Justice Department over allegations they failed to prevent roughly 80,000 illegal drug and chemical sales through Alibaba’s platforms between 2016 and 2024.
Britain’s Financial Conduct Authority has announced a comprehensive regulatory framework for crypto firms, requiring capital holdings against risky assets and annual stress tests, coming into force in October 2027. Binance founder Changpeng Zhao said the exchange’s EU MiCA licence application was fully compliant and near approval in Greece before unspecified political intervention led to its withdrawal, while separately almost 1,700 British investors have filed a £150 million lawsuit against Binance in London’s High Court alleging it sold risky crypto derivatives to UK retail investors without authorisation.
A US Supreme Court ruling strengthening presidential control over independent agencies has renewed European concerns about the stability of the EU-US Data Privacy Framework. Privacy activist Max Schrems has used the ruling to challenge the framework, arguing it undermines the independence of the US Federal Trade Commission, a key oversight body under the deal underpinning an estimated €1.7 trillion in annual EU-US trade.
Australia’s Fair Work Commission says AI-assisted litigation is driving a record increase in dismissal-related claims, with total cases expected to exceed 55,000 this financial year after rising more than 70% over three years. Commission president Justice Adam Hatcher estimates more than half of applicants now use AI to prepare claims.
In the UK, AI tools are similarly making it cheaper for employees to assess employment claims, according to TechRound, contributing to a doubling of the tribunal case backlog in two years.
⸻
💰 Tech business & markets
Australia’s data centre sector, valued at about US$97 billion with applications for an additional 5.4GW of capacity, may deliver limited domestic economic benefit because most equipment is imported and ownership is largely foreign, economists including HSBC’s Paul Bloxham have warned.
South Korea’s government and chipmakers Samsung Electronics and SK hynix will invest 800 trillion won (US$519 billion) to build a semiconductor complex in the country’s southwest. The plan also includes AI data centres backed by SK Group, GS Group and Naver, with an initial capacity of 8.4GW.
In Florida, growing local opposition is causing developers to reconsider large-scale data centre investments, with Citrus County imposing moratoriums and Fort Meade residents suing over a proposed Stonebridge development.
Intel is showing signs of recovery, with its market value tripling and renewed demand for its data centre chips driven by the AI boom, though the company continues to lose money. In South Korea, record chip exports and bonuses at Samsung and SK hynix have sharply increased interest in vocational schools training students for semiconductor careers, though experts warn manufacturing creates relatively few jobs.
New data from the Stanford Digital Economy Lab and ADP Research indicate employment for workers aged 22-25 in highly AI-exposed occupations is shrinking by 3.8% annually, while less-exposed jobs for the same age group continue to grow. In China, rumours of major job cuts at food delivery company Meituan have heightened anxiety among tech workers about AI-driven restructuring, though Meituan denied reports it planned to cut up to half of its product roles. In Germany, SAP says it is using AI to transform jobs rather than eliminate them, despite cutting nearly 10,000 positions during a restructuring partly linked to AI adoption, and has since added more than 3,500 roles.
AI is increasing productivity across Silicon Valley but also driving burnout and fear of obsolescence among founders and engineers, Bloomberg reports, as companies push staff to demonstrate productivity comparable to AI agents.
Silicon Valley figures and AI companies are pledging hundreds of millions of dollars to support candidates aligned with their preferred AI regulatory approach, with AI also reshaping campaigning through generative advertising and deepfakes ahead of the 2028 presidential race. The New York Times separately reports that US political campaigns are increasingly using AI to analyse voter data and tailor messages ahead of the November midterms.
Indian IT services firms are increasing acquisitions as AI reshapes the sector’s growth prospects and pressures traditional outsourcing models, with a greater willingness to use debt financing for larger deals.
Travel companies including Accor and Skyscanner are investing in conversational interfaces and direct booking capabilities to maintain customer relationships as autonomous AI agents begin planning and booking travel. In aviation, the UK’s National Air Traffic Services is testing AI through Project Bluebird to model traffic scenarios and assist controllers, though industry experts say fully replacing human controllers is unlikely.
Uncertainty over US immigration policy is prompting skilled technology workers to consider relocating to Canada, the UK, the EU and the Gulf, with H-1B registrations for fiscal 2027 down 38.5% year on year despite a court ruling invalidating a US$100,000 fee on new H-1B visas.
Brazilian football clubs including Santos F.C. are adopting AI-powered scouting apps such as Cuju and Footbao to broaden talent identification beyond traditional networks, though critics warn of bias and access barriers. Elsewhere, Japanese startups including Miraima are building points-based prediction markets that mimic the legal structure of the pachinko industry to sidestep restrictions on real-money gambling, reportedly reaching one million monthly users in seven months.
⸻
🌏 Rest of the world
Australia
An analysis of 14 large pro-One Nation Facebook groups found most were created this year and appear to be administered by overseas digital creators, primarily in Indonesia and India, who monetise engagement through Meta’s creator programs using AI-generated and outrage-driven content. Meta removed at least one impersonation account after being contacted by Guardian Australia and said it was reviewing the content.
The Reserve Bank and Transport for NSW have raised concerns that rapid data centre investment could worsen inflation and construction labour shortages and crowd out industrial land needed for housing and freight.
Europe
French President Emmanuel Macron has instructed officials to identify new EU-wide taxes to help finance the bloc’s next €2 trillion seven-year budget, pushing for levies on digital giants, foreign polluters, crypto firms and online gambling. Tax measures require unanimous backing from all 27 member states, and the Commission’s earlier revenue package has already faced resistance.
The European Commission has rejected US criticism of EU technology regulation and warned it would respond ‘swiftly and decisively’ to any tariffs targeting European digital policies, amid rising tensions over digital service taxes and EU efforts to reduce dependence on US technology providers.
India
India has launched a hackathon to develop affordable, multilingual AI devices that work offline and run on open-source models, with government-backed platform Bhashini and French nonprofit Current AI supporting teams building tools for education, agriculture and healthcare in areas with limited connectivity.
New Zealand
ACT Party candidate Lyra Yan Zhang resigned from New Zealand’s Kenepuru electorate race after failing to disclose her previous membership of the China Zhi Gong Party, a body linked to the Chinese Communist Party’s United Front Work Department. The case emerged amid intelligence agency warnings that China remains the most active state actor conducting interference activities in New Zealand.
Singapore
The Monetary Authority of Singapore plans to establish a Future of Finance Institute to accelerate adoption of AI and tokenisation across the financial sector, building on existing programmes including PathFin.ai, Project Guardian and Project Orchid.
That’s all for this week. For more timely analysis and commentary, check out The Strategist and ASPI’s Stop the World podcast—or our other Substack newsletters:
The Cyber & Tech Digest is brought to you by the Cyber, Technology & Security Programs team at ASPI and supported by partners.
