0:09
Anthropic deploys Claude Sonnet 5, Fable and Mythos restored
Anthropic has launched Claude Sonnet 5 and restored access to its Fable and Mythos frontier models following a federal export control review.
The decision marks the conclusion of an eighteen-day operational pause triggered by a US government export control directive on June 12, which forced the temporary suspension of Anthropic’s highest-capability systems.
Government officials enacted the restriction after researchers at Amazon documented a method to bypass the safety controls of Fable 5, causing the model to identify software vulnerabilities and supply exploitation code. Anthropic has since developed an updated automated classifier to patch the vulnerability, clearing the path for a full commercial rollout across its platform, cloud infrastructure, and partner networks.
The temporary suspension of Fable 5 and Mythos 5 highlighted the regulatory pressures facing frontier intelligence systems. When the export control mandate took effect, the lack of real-time nationality verification systems required a total access blackout for all global users.
Security evaluations conducted during the shutdown confirmed that the vulnerability identification behaviour was not unique to Fable 5. Older and less capable architectures from multiple providers, including Claude Opus 4.8, GPT-5.5, and Kimi K2.7, duplicated the exact results.
To resolve the federal directive, engineers trained an automated safety classifier targeting the specific bypass mechanism reported by Amazon. This software layer functions with a wide safety margin, identifying and blocking ambiguous developer prompts that display a statistical probability of malicious intent. Internal validation data indicates the updated classifier prevents the reported exploitation technique in more than 99 percent of trials.
When a developer issues a prompt that triggers this boundary, the platform automatically routes the workload to the older Opus 4.8 architecture to maintain continuity. The expanded safety margin introduces a distinct trade-off for engineering teams, as the automated system flags benign requests more frequently during routine application development and software debugging.
While frontier models face strict state oversight, the immediate commercial focus targets the newly-deployed Claude Sonnet 5.
Engineering teams are transitioning autonomous agents to this model to reduce operational expenditure while maintaining high execution capacity. Performance data validates that the system executes multi-step plans, operates terminal environments, and navigates web browsers without human intervention.
Model performance and cost metrics:
*Cost per million tokens. Sonnet 5 carries introductory rates of $2.00 input / $10.00 output through August 31, 2026.
Real-world deployments demonstrate how organisations are deploying this architecture within live software development pipelines.
At Rakuten, technology teams deployed the architecture against dozens of the company’s most challenging production code pull requests. The system processed each submission independently, executing tests and verifying the results before presenting the completed code to human engineers for final structural approval.
Software automation firm Zapier integrated the system into its core product workflows to execute multi-part administrative tasks. In a documented deployment, engineers tasked the model with updating Salesforce account tiers and subsequently generating and transmitting launch announcements to enterprise contacts. Prior model architectures frequently stalled midway through these multi-stage operations, whereas the current system executed the entire sequence end-to-end without human remediation.
Development tool provider Zed utilised the system to automate complex debugging procedures. During internal trials, engineering teams directed the model to investigate an active software bug. Working without explicit prompts or step-by-step instructions, the system independently generated a reproducing test script, applied the necessary code fix, and stashed the modifications to verify that the bug reappeared in the absence of the patch. The entire diagnostic and remediation sequence occurred within a single processing pass.
Software engineering platform Factory implemented the architecture to manage sustained coding tasks within complex codebase environments. Technical teams reported that the system maintained logical grounding and execution consistency across corporate code repositories, outperforming previous generation software layers by completing tasks that previously timed out or failed to resolve.
Data from the formal system card indicates that the system achieves these autonomous capabilities without a corresponding inflation of security risks.
5:01
Bank of England reviews AI rules for agentic AI in finance
The Bank of England is reviewing whether existing rules can cover the use of agentic AI in finance, including payments, trading, cybersecurity, and operations.
Deputy Governor Sarah Breeden said existing regulatory frameworks were not designed for AI agents that can act without direct human instruction. Speaking at the European Central Bank Forum on central banking in Portugal, she said relying on human oversight for every action by these systems is unlikely to be practical.
Breeden said current frameworks were not built to contemplate autonomous agents in payments, trading, and operational functions.
Agentic AI refers to systems that can make decisions and carry out tasks independently. In finance, such systems are already being used in areas such as product recommendations, operational workflows, and trading-related tasks.
Agentic systems differ from traditional automated trading tools because they can pursue objectives and make decisions with less direct human supervision. Breeden said these systems could act in similar ways if they are trained on similar data or designed around similar goals.
Breeden said recent advances in AI models for identifying cyber vulnerabilities show a change in capability. She said agentic AI systems can chain together sequences of actions at scale and speed.
A 2026 Cambridge Centre for Alternative Finance report found that 81% of surveyed financial services firms are adopting AI at some level. It also found that 52% of industry respondents are already actively adopting agentic AI.
The report said most current use remains focused on internal functions, including process automation, data visualisation, software engineering, and knowledge management. Breeden said use in trading is still mostly concentrated in lower-risk operational tasks.
Breeden described cyber resilience as one of the Bank of England’s closest financial stability concerns around agentic AI. She said the technology has undergone a “step change” in cyber capability and that supervisors need to look at risks across the financial system rather than only at individual firms.
She said AI tools can strengthen cyber defences when used by security teams. The immediate risk, she added, is that the same tools could increase the chance of attacks that harm financial stability if used by malicious actors.
Breeden also noted that open-source models may trail the most advanced closed models by only four to eight months. She said this gives authorities only limited comfort, despite restrictions on the release of some advanced models.
The IMF has also warned that AI-enabled cyber risk should be treated as a financial stability issue. It said attacks can scale quickly, spread across sectors that share digital infrastructure, and create wider disruption if several institutions are affected at once.
Breeden said authorities should place greater weight on simultaneous disruption across several firms and stress-test the likely impact before such events occur. She said recovery planning may also need to account for mass disruption, rather than only isolated outages.
The Bank of England is considering stronger recovery requirements for core systems. One option is to allow one bank to take over another bank’s basic functions during an outage or failure.
Other options include arrangements that allow critical services to continue if a firm’s core systems are compromised. Breeden also raised the question of whether key firms should have separate failover systems or the ability to rebuild compromised core systems quickly.
Tobias Adrian, financial counsellor and director of the International Monetary Fund’s capital markets department, also said AI poses serious risks to cyber resilience, according to Central Banking. The IMF has separately warned that shared software, cloud services, payment networks, and data networks can create correlated failures if widely used systems are targeted.
Breeden said regulators are also looking at guardrails, circuit breakers, and kill switches. These tools would be designed to limit or stop trading across markets if faulty AI models contribute to severe disruption.
Breeden said autonomous systems could amplify volatility if they respond in similar ways to the same market signals, especially if their objectives drift from their original purpose or from public policy goals.
The Bank of England has previously said existing rules were sufficient to manage AI-related risks. Breeden said recent developments have exposed gaps in current frameworks.
The Financial Stability Board said earlier in June that AI agents pose a distinct challenge for human oversight and called for stronger safeguards.
The FSB’s June consultation set out 12 proposed sound practices for responsible AI adoption by financial institutions.
9:56
Japan’s answer to its worker shortage: An AI model for 10 million robots
Japan’s AI robots plan just went from a talking point to a formal national strategy. This week, the government confirmed the numbers everyone’s been quoting: 10 million AI-powered robots deployed across 18 industries by 2040, backed by public funding of up to one trillion yen, or roughly US$6.1 billion, over five years.
The headline figure is the kind that gets shared without much scrutiny. What’s easy to miss is that this isn’t a policy wish list either. It’s a project the government has now formally commissioned, and the company doing the building is one most people outside Japan haven’t heard of.
METI and NEDO, Japan’s industry ministry and its innovation agency, have formally commissioned Noetra and AIST, a national research lab, to develop a “physical AI” model as part of a push running from fiscal 2026 to 2030. The goal is a multimodal foundation model, one that can read language, images, video and sensor data together, so a robot can actually interpret a room and act in it rather than just execute pre-programmed motions.
An initial version is due out as early as this fiscal year, with annual upgrades after that, built using data volunteered by manufacturers and other participating companies. The money isn’t unconditional, either. The current fiscal year’s commission is reportedly worth around US$2.3 billion on its own, drawn from a 387.3 billion yen allocation funded through GX Economy Transition Bonds.
Only the first two years are locked in. After that, funding gets reviewed annually through a stage-gate process, meaning Tokyo can pull back if Noetra misses its milestones. For a project this size, that’s a meaningful detail: the trillion-yen figure is a ceiling, not a guarantee.
Noetra is majority-owned by SoftBank, NEC, Sony Group and Honda, with Fujitsu and Rakuten reportedly weighing whether to join. SoftBank engineers are working alongside researchers from Preferred Networks and AIST itself.
It’s a familiar shape for a Japanese industrial push: rather than one company chasing a frontier model alone, the state has assembled a consortium of firms that already build the hardware this model needs to run on, from Honda’s robotics to Sony’s imaging sensors.
Industry minister Ryosei Akazawa has been direct about the reasoning. The plan, he said, will “vigorously promote social implementation” across sectors, including restaurants, food manufacturing and medical care. Behind that language is a labour market running out of people: Japan’s ageing population, combined with tight migration policy, has left large parts of the economy short of workers with no easy fix in sight.
Japan isn’t starting from nothing here. The country has spent years building robotics expertise in elder care, disaster response, manufacturing and even the Fukushima Daiichi cleanup. This project is an attempt to turn that experience into something exportable, not just a domestic patch.
The timing also isn’t a coincidence. South Korea announced its own robotics push within a day of Japan’s confirmation, and both governments are framing physical AI as the next front in a competition that’s mostly been fought over chatbots and cloud contracts until now.
The real test isn’t the 2040 target, it’s the first stage-gate review. If Noetra hits its early milestones and releases a usable model this fiscal year, expect the investor list to grow well beyond the current four. If it doesn’t, the funding structure gives Tokyo every reason to walk away quietly rather than prop up a stalled national project.
See also: From cloud to factory – humanoid robots coming to workplaces
Want to learn more about AI and big data from industry leaders? Check out AI & Big Data Expo taking place in Amsterdam, California, and London. The comprehensive event is part of TechEx and is co-located with other leading technology events, click here for more information.
AI News is powered by TechForge Media. Explore other upcoming enterprise technology events and webinars here.
The post Japan’s answer to its worker shortage: An AI model for 10 million robots appeared first on AI News.
14:09
Anthropic's Fable 5 is back worldwide after a two-week government ban over a jailbreak
After a two-week ban, the US government is letting Anthropic ship Fable 5 globally again. Amazon researchers had found a jailbreak, but Anthropic says even much smaller models like Claude Haiku 4.5 could pull off the same exploit. A new safety classifier blocks the technique in over 99 percent of cases, though it also flags more harmless requests in the process.
The article Anthropic's Fable 5 is back worldwide after a two-week government ban over a jailbreak appeared first on The Decoder.
14:47
OpenAI paper reveals three GPT-5.6 Pro models, breaking with single top-tier strategy
An OpenAI benchmark paper suggests that the Pro tier of GPT-5.6 could ship in three variants. That would be the first major change to ChatGPT Pro's structure since the plan launched.
The article OpenAI paper reveals three GPT-5.6 Pro models, breaking with single top-tier strategy appeared first on The Decoder.
15:13
Claude Sonnet 5 continues Anthropic's pattern of hiding price increases behind unchanged token rates
Claude’s newest model, Sonnet 5, looks a lot like its predecessor on paper—same token price, same headline specs. The twist is that under the hood it’s chewing through roughly 40 percent more tokens per task, which means the real bill is almost double what you’d expect from the listed rate.
That extra token appetite shows up in the AI Index, where Sonnet 5 lands at 53 points, nudging into fifth place and even edging out the pricier Opus 4.8 on a handful of agent‑based benchmarks. It’s a neat performance bump, but the cost side‑effect is what’s catching attention.
Anthropic’s pattern is becoming clearer: they keep the per‑token price static while the model’s internal token consumption climbs, effectively masking a price hike. It’s not a new headline, but it does shift the economics for anyone budgeting on usage.
If you were counting on the old token price to keep expenses predictable, you’ll need to factor in that hidden increase now, especially if you’re running many tasks that hit the higher token counts.
16:23
Hidden code in Claude Code secretly flagged Chinese users
Anthropic is removing a hidden monitoring feature from its programming tool, Claude Code, after it sparked outrage on social media.
The article Hidden code in Claude Code secretly flagged Chinese users appeared first on The Decoder.
16:43
Claude Science is Anthropic’s newest flagship product
At an event for pharmaceutical executives, biotech founders, and researchers on Tuesday, Anthropic announced Claude Science, a major new product intended to support scientific research in the same way that Claude Code supports software engineering. Like Claude Code, Claude Science can autonomously carry out meaningful work when given concise, high-level instructions, and it has access to tools that make it particularly useful for research in computational biology and drug development. Along with launching and previewing Claude Science, which is now available to all paid Claude subscribers, Anthropic also announced that it will be using the product to pursue some of its own research into drugs for rare, neglected diseases.
This is not Anthropic’s first foray into AI for science. In October, the company released plug-ins that help Claude make use of scientific software and databases under the heading “Claude for Life Sciences.” But unlike this earlier release, Claude Science is a full-featured, standalone product. Anthropic’s decision to elevate Claude Science to the same rank as Claude Code and Claude Cowork indicates that the company is taking AI’s scientific applications very seriously—or at least wants to give the impression that it is.
“It represents how important this is to our mission that this is right up there with Claude Code and Claude Cowork as the next really significant product that we’re releasing,” says Eric Kauderer-Abrams, Anthropic’s head of life sciences. “Our mission is to develop AI that serves humanity’s long-term well-being, and we believe that by far the greatest opportunity to do that is in the life sciences.”
For the past decade, one company—Google DeepMind—has been at the vanguard of AI for science. CEO Demis Hassabis and researcher John Jumper won the Nobel Prize in chemistry for their work on the company’s AlphaFold model, and DeepMind has also made major contributions to meteorology, materials science, and a variety of other disciplines. But in the past several months, the fast-advancing frontier of AI progress seems to have left DeepMind in the dust. When it comes to coding, which has become the most lucrative use case for LLMs, DeepMind is stuck playing catch-up.
Anthropic is well positioned to take up DeepMind’s scientific mantle. Like Hassabis, Anthropic CEO Dario Amodei is a PhD scientist—unlike OpenAI CEO Sam Altman, who’s a businessman through and through. Many scientists are already avid users of tools such as Claude Code. These days, a lot of scientific research involves some amount of coding, but not all scientists are expert software engineers, and so tools like Claude Code can make a huge difference for their productivity. And the company has recently earned a major scientific vote of confidence: Earlier this month, Jumper announced that he is leaving DeepMind for Anthropic.
Since agents powered by LLMs, including Anthropic’s Opus model series, became capable of useful, independent work in late 2025, scientists have been seeing just how much they can do. In a blog post published on Anthropic’s website, the Harvard physicist Matthew Schwartz estimated, on the basis of his work with Claude Code and other Anthropic tools, that the company’s Opus 4.5 model is about as capable of executing scientific projects as a second-year graduate student.
According to Kauderer-Abrams, Claude Science isn’t intended to displace Claude Code and Claude Cowork in scientists’ workflows. Instead, it’s designed to build on what scientists already find useful about Anthropic’s products. For instance, it not only writes code but also helps scientists run their code on powerful computer clusters, which many many scientists need for their work but can be difficult to manage. And it prioritizes reproducibility, so that scientists can trace back the source of any figure or result and check it for accuracy and validity.
Though Claude Science could in principle assist with any area of scientific research, it seems designed and marketed as a tool for molecular and cellular biology, and for drug development in particular. It can interface with various tools used in genetics, chemistry, and protein biology, all of which could come in handy for researchers on the hunt for new drugs. During the Tuesday event, Alexander Tarashansky, who led the development of Claude Science, demonstrated how the system could autonomously identify new drug candidates for phenylketonuria, a rare genetic disease.
And Anthropic isn’t leaving all that work to the pharma companies and university labs that were represented at the event.