The US government has lifted export controls on Anthropic’s Mythos and Fable artificial intelligence models, prompting a wider debate among vendors and policymakers over the resilience and security of AI infrastructure.

The initial restriction, which limited foreign access to some of Anthropic’s most advanced systems, exposed how quickly geopolitical or regulatory decisions can disrupt the supply of closed foundation models. AI founders and government technology advisers argue the episode will have lasting effects on how states and enterprises structure their AI stacks, even as access resumes.

Eugene Cheah, Chief Executive Officer and Co-Founder of Featherless.ai, said the episode had already shifted sentiment toward open-source alternatives that organisations can deploy on their own infrastructure.

“This is welcome news from Anthropic. However, I would be concerned that the damage has already been done. It showcased how fragile access to closed models really is and may push more governments, companies, and developers toward open models they can run, adapt, and keep access to themselves.

“That is the point of open source: research is not limited to a small number of labs, and teams can tune models for the problems they are trying to solve. Rio 3.5 Open, a model released by the City of Rio de Janeiro, is a good example. It shows how far open-model work can be pushed when it is adapted for a specific use case,” he said.

Featherless.ai, which recently closed a USD $20 million Series A round led by AMD Ventures and Airbus Ventures, focuses on infrastructure that allows enterprises to run specialist models across varied hardware in their own jurisdictions. Cheah said this approach reduces reliance on a small group of hyperscale providers and gives organisations greater control over where and how AI systems operate.

He expects closed models to retain an edge on the most complex workloads, but said most businesses do not need premium-grade systems for every task.

“Closed models will still be ahead in some areas, especially on the hardest and most expensive tasks. But most organisations do not need a Porsche for every journey. They need models that are good enough for the job, that they can keep access to, and that are not dependent on decisions made by a handful of companies,” Cheah said.

Anthropic has also begun redeploying Claude Fable 5, a version of its Mythos tool that it had previously judged too risky for broad release. The system includes new safeguards and user limits, and Anthropic has acknowledged that “releasing a model this capable comes with risks”.

Roman Stanek, Founder and Chief Executive Officer of GoodData.AI, said longstanding failures in basic cybersecurity remain the bigger issue around such advanced models.

“The thing with AI and cybersecurity is that the vulnerabilities we’re worried about AI exploiting, well, we’ve known about most of them for 20 years. We just never fixed them. Not because we couldn’t, but because nobody wanted to pay for them. Open-source security, legacy code debt, infrastructure hygiene – they are all solvable problems, but all chronically underfunded. Now we have Mythos, this newer, shinier, more expensive model that can theoretically find and fix them.

“Great. Except the incentive structure hasn’t changed. Nobody wanted to pay a human engineer to fix it. They’re not going to pay an AI to fix it either. The issue has never been capability. People are just unwilling to invest, and until that changes, the vulnerabilities will remain as the threat evolves. We will continue to have the same conversation around better and better tools that nobody will use,” Stanek said.

For public authorities, the incident has intensified discussion of “AI sovereignty”. John Harms, Head of Government Solutions at Quantexa, said it underlined the importance of control over the data and decision-making layers that sit above any particular model.

“Government IT leaders regularly tell me that uncertainty is a significant concern this year. Recent events surrounding access to Anthropic’s Mythos and Fable models illustrate why. Even though the decision affecting foreign access has been reversed, the episode serves as a reminder that organisations can be exposed to risks outside their direct control.

“This is not a US-versus-the-rest-of-the-world issue. Any government or organisation, regardless of geography, needs confidence that the technologies underpinning critical operations will remain available. The lesson is clear: resilience depends on sovereignty. And sovereignty requires ‘control under distress’,” Harms said.

He defined that as the ability to sustain operations through geopolitical, regulatory, commercial, or technical disruption while retaining authority over data, models, and the software that turns outputs into real-world decisions.

“Control under distress is the ability to continue operating confidently during geopolitical, regulatory, commercial, or operational disruption. It means being able to govern data, AI models, decision-making processes, infrastructure, and operational dependencies under all circumstances,” Harms said.

He said governments do not need to insist on purely domestic models, but must avoid structures in which external “kill switches” can halt critical services.

“This doesn’t mean we should reject all global technology and insist that every model is built in the UK. That would limit choice and weaken competition. But organisations need to be in a position where they still control their data and decision intelligence layer for critical aspects of the tech stack.

“The prospect of externally controlled ‘kill switches’ raises understandable concerns. Public sector leaders should look at every level of their technology stack, from infrastructure and models all the way to the software on which decisions are being made,” Harms said.

SOFIAH NICHOLE SALIVIO

News Editor

PFU has launched the Ricoh SP-2240N and Ricoh SP-2230N document scanners for routine business scanning tasks, priced at GBP £531 and GBP £416.

Based on Ricoh fi Series technology, the models are designed for day-to-day document digitisation. The SP-2240N scans at up to 40 pages per minute, while the SP-2230N reaches up to 30 pages per minute.

Both scanners are built for shared office use and support a range of document types, including standard business papers, plastic cards and A3 documents folded in half. Each has an 80-sheet automatic document feeder in a compact desktop format.

Shared Scanning

A key difference between the models is the SP-2240N’s support for networked and PC-free scanning through DirectScan. The feature is enabled by PFU’s new iiGA system on a chip, allowing users to scan directly from the device and send files to preset destinations without needing a connected computer for each job.

The scanners include three on-device buttons that can be assigned to functions such as saving files to set locations or sending them to email recipients. A PC is still required to configure scanner settings.

Brian Fortune, General Manager Sales, PFU EMEA, said the higher-end model was designed for organisations with constrained technology resources.

“The RICOH SP-2240N represents what many organisations are looking for today: a practical and reliable scanning solution that balances performance, ease of use and cost. As businesses work to digitise documents with limited IT resources, the SP-2240N delivers consistent, high-quality results across shared environments, with the flexibility to process scans directly on the device when a PC is not available,” Fortune said.

Image Processing

The new models use Clear Image Capture technology and include optical character recognition and image processing tools intended to reduce manual correction work. The software can handle functions such as paper size detection and automatic rotation to improve output accuracy across different document formats and orientations.

The scanners also include a Paper Protection function with real-time image monitoring to help prevent document damage during regular use. This is intended to support uninterrupted operation in offices with frequent scanning workloads.

For PC-based workflows, the scanners come with PaperStream ClickScan software, allowing users to start digitisation from a right-click scan command or by pressing a button on the scanner itself.

The package also includes PaperStream Capture software as standard. Customers needing additional automation or form capture functions can upgrade to Capture Pro, while PaperStream IP Net allows any networked PC and scanner to work together.

Office Demand

The launch comes as vendors continue to position document scanners for small and mid-sized office digitisation, where paper records remain common and IT support can be limited. In this market, compact size, lower upfront pricing and simpler setup are often key buying considerations.

PFU said the SP-2240N’s chip also supports the creation of searchable PDF files during PC-free scanning, presenting this as part of a broader push towards one-touch workflows in multi-user office settings.

Fortune said: “With our next-generation iiGA System on a Chip, we are delivering unmatched OCR results for PC-free scanning. Users can now instantly create searchable PDFs and send documents to pre-programmed destinations with the simple press of a button, allowing teams to maintain high productivity without the need for a PC.

“These scanners also provide the flexibility to grow with a business. They come supplied with PaperStream Capture software, which is optimised to deliver the best possible performance for scanning routines. For organisations that need more, an upgrade to Capture Pro offers powerful automation and intelligent form capture, even for handwritten non-cursive text. Furthermore, the inclusion of PaperStream IP Net ensures that any networked PC and scanner can be used together, supporting business scalability from the very first day.”

PFU, founded in Japan in 1960, focuses on document imaging products and IT infrastructure services. It said it holds the top global market share in document scanners, citing market research covering Japan, North America and Western Europe.

SOFIAH NICHOLE SALIVIO

News Editor

Jitterbit has published research showing a divide between UK and US technology leaders over whether to buy or build AI agents, highlighting a split in enterprise preferences for agentic automation.

Its data shows that 51% of UK organisations prefer pre-built agents for building and deploying AI agent automations, while 48% rely on custom-made systems. In the US, the pattern is reversed: 55% of respondents favour custom-built code developed entirely in-house, while 45% lean toward pre-built agents.

The figures come from Jitterbit’s 2026 AI Automation Benchmark Report, which examines how IT decision-makers are approaching AI and automation. It suggests UK companies place greater value on speed and efficiency than on bespoke development as they expand their use of AI agents.

Among UK respondents, 43% preferred agents embedded in pre-built software-as-a-service applications. By contrast, 28% preferred building their own low-code or no-code integration platforms, while 20% favoured in-house custom-coded agents built through tools such as Lovable or similar services. Only 8% chose off-the-shelf agents as their primary source.

Cost pressure

Jitterbit linked the strategic divide to the changing economics of AI deployment, arguing that rising token costs and heavier use of large language models are making bespoke development less attractive, particularly for companies weighing the ongoing cost of maintaining internal systems.

This reflects a wider market concern over the total cost of AI projects. Many companies have invested heavily in internal engineering teams and AI experimentation, but spending on inference, integration and model usage has become a more visible budget issue as projects move into broader business use.

The report also found that 81% of organisations plan to increase funding for AI and automation over the next 12 months, suggesting demand remains strong even as businesses become more selective about deployment and expected returns.

Build or buy

The contrast between the UK and the US also points to different corporate cultures around software development. US engineering teams appear more willing to keep AI agent development in-house, while British organisations show greater readiness to adopt pre-built options from software vendors.

The distinction matters because AI agents are moving from pilot projects into mainstream operational tools. Companies are using them for process automation, application integration and task orchestration, so decisions on whether to build internally or buy ready-made systems can affect budgets, deployment times and the level of technical oversight required.

Pre-built systems can offer faster implementation and more predictable supplier support, while custom builds can provide greater control over workflows and internal data handling. The findings suggest this trade-off is becoming sharper as AI consumption costs rise and boards demand clearer evidence of value.

For UK businesses, the research indicates that caution over custom development is not simply technical conservatism. It also reflects a pragmatic response to the pressure of deploying AI tools at scale without taking on open-ended costs tied to internal development and token use.

Bill Conner, president and chief executive officer of Jitterbit, said the market is moving into a new phase as AI projects become more integrated into core operations.

“We are moving past the era of siloed AI experimentation and entering the era of pure orchestration,” Conner said.

He said economic pressure is reshaping how companies assess custom systems against pre-built alternatives.

“The market is waking up to the true total cost of ownership of custom AI. Between developer bottlenecks and soaring token costs, building everything from scratch is a luxury few can sustain. The UK’s slight preference for pre-built applications may have previously been considered cautious, but it’s proving to be highly strategic as economic pressures force a shift from development to intelligent integration,” Conner said.