Author: Matthew Miller

Brief Overview

• The US authorities have announced an export control order that impacts Anthropic’s Claude Fable 5 and Mythos 5 models.
• Access to these models is halted for all non-US individuals, including foreign Anthropic employees.
• The order is driven by worries about a potential jailbreak vulnerability in Claude Fable 5.
• Though Anthropic disagrees with the decision to recall, it is complying with the order to maintain security.
• Mythos 5 was launched under restricted conditions, featuring fewer protections compared to Claude Fable 5.

Effects of the US Export Control Order

In a notable action, the United States government has enacted an export control order mandating Anthropic to suspend access to its Claude Fable 5 and Mythos 5 models. This order influences all non-US nationals, including foreign employees of Anthropic, both in the US and abroad. The repercussions are extensive, impacting users of these models globally.

Security Issues and Adherence

Anthropic has stated that the action taken by the US government is premised on concerns regarding a possible jailbreak vulnerability in Claude Fable 5. The company characterized the vulnerability as “a potential narrow, non-universal jailbreak,” which entails directing the model to evaluate a specific codebase and correct software defects. Despite this, Anthropic has conveyed its disagreement with retracting a model already utilized by hundreds of millions worldwide.

Launch and Protections

Claude Fable 5, which became publicly available on June 10, was fitted with measures to mitigate misuse in the spheres of cybersecurity and biological and chemical research. In contrast, Mythos 5 was disseminated to chosen partners under Project Glasswing and lacked certain layers of protection present in Claude Fable 5, although both models utilize the same core technology.

Anthropic’s Reaction and Prospective View

Although not having received a formal notification of a non-universal potential jailbreak resulting in adverse effects, Anthropic has examined a report believed to underlie the US government’s prohibition. The company, in association with the US government and the UK’s AI Security Institute (AISI), conducted comprehensive evaluations of the models’ protections before releasing Claude Fable 5.

Anthropic upholds a 30-day data retention policy to track usage and has embraced a defence-in-depth approach to make jailbreaks difficult and expensive to achieve. Nonetheless, the company recognizes the challenges in achieving complete jailbreak resistance and expects to uncover minor circumventions of protections and potentially more extensive jailbreaks in the future.

Conclusion

The sudden suspension of Anthropic’s Claude Fable 5 and Mythos 5 models following a US export control order underscores the ongoing difficulties in AI model security and regulatory compliance. While Anthropic seeks to address the raised issues, the company continues to emphasize security and protective measures in its models.

Q: Which models are impacted by the US export control order?

A: The impacted models are Anthropic’s Claude Fable 5 and Mythos 5.

Q: What prompted the US government to issue the order?

A: The order is driven by concerns regarding a potential jailbreak vulnerability in the Claude Fable 5 model.

Q: How does this order affect Anthropic employees?

A: The order also pertains to non-American Anthropic employees, hindering their access to these models.

Q: What distinguishes Claude Fable 5 from Mythos 5?

A: Claude Fable 5 includes protections for cybersecurity and research, whereas Mythos 5, released under Project Glasswing, has fewer protections but shares the same foundational technology.

Q: What security protocols did Anthropic establish for Claude Fable 5?

A: Anthropic established a defence-in-depth strategy and a 30-day data retention policy to oversee model usage and complicate jailbreak attempts.

Q: Are there known successful jailbreaks of these models?

A: Anthropic has not received any official notice of successful non-universal jailbreaks that resulted in negative consequences for Claude Fable 5.

Q: What is the prospective outlook for the security of Anthropic’s models?

A: Anthropic anticipates possible minor circumventions of protections but is striving to avert widespread jailbreaks while recognizing the challenges in achieving flawless resistance.

Brief Overview

• Apple announces new OS enhancements at WWDC, emphasizing AI and security improvements.
• The Passwords app now includes automated credential updates powered by Apple Intelligence.
• Enhanced network security with TLS 1.3 recommended for all servers.
• macOS 27 will cease support for Intel-based Macs, requiring Apple Silicon chips.
• Upgraded child safety features in iOS 27, featuring content filters and parental controls.

AI Becomes a Focus at WWDC

During the Worldwide Developers Conference (WWDC), Apple introduced innovative updates to its operating systems, with artificial intelligence (AI) prominently featured. The new version 27 OSs for Apple devices will incorporate on-device foundation models from Google, utilizing privacy-focused Private Cloud Compute technologies.

Password App Improvements

Among the notable security enhancements is the revamped Passwords app. This app utilizes Apple’s sophisticated AI to detect weak, reused, and compromised passwords. It can automatically update passwords on websites and store the new credentials on the user’s device. However, its efficiency with sites that require multi-factor authentication and CAPTCHA codes is yet to be determined.

Network Security Advancements

Apple has strengthened network security by recommending Transport Layer Security (TLS) version 1.3 for servers, setting 1.2 as the minimum threshold. The App Transport Security (ATS) function will prohibit unsecured connections that lack TLS protection. These modifications will affect mobile device management, software updates, and app installations, including enterprise app distribution.

macOS 27 Compatibility Changes

With the introduction of macOS 27, support for Intel-based Macs will be phased out, mandating the use of Apple Silicon chips. Supported devices will encompass Mac mini, MacBook Pros, and Air starting from 2020, iMacs from 2021, and newer models. Furthermore, the full AI functionalities of macOS 27 necessitate an M3 chipset and a minimum of 12GB of memory, restricting some older models’ capabilities.

Upgraded Child Safety Features

Apple has enhanced child safety features within its latest OS updates. The Communication Safety feature now filters out violent content in Messages and FaceTime for users under the age of 18. A redesigned Screen Time limits feature, a suggested app set, and the ‘Ask To Browse’ functionality in Safari improve parental controls. Additionally, developers can utilize the Declared Age Range API to ask for age ranges without requiring actual birthdates.

Conclusion

Apple’s recent announcements at WWDC emphasize major strides in security and AI integration within its operating systems. With new capabilities such as automated password updates, stricter network security protocols, and enhanced child safety initiatives, Apple continues to establish high benchmarks in digital security and user experience.

Q&A

Q: What are the primary security updates in Apple’s new OS?

A: The primary security updates encompass an upgraded Passwords app for automatic credential updates, enhanced network security with TLS 1.3, and improved child safety features.

Q: Will older Macs remain compatible with macOS 27?

A: No, macOS 27 will exclusively support devices equipped with Apple Silicon chips, phasing out Intel-based Macs.

Q: How does the upgraded Passwords app function?

A: The upgraded Passwords app leverages Apple Intelligence to recognize and automatically update weak or compromised passwords, saving the new credentials on the user’s device.

Q: What enhancements are being implemented for child safety?

A: Enhancements consist of content filtering for Messages and FaceTime, a revamped Screen Time limits feature, and the ‘Ask To Browse’ functionality in Safari for better parental controls.