2 weeks ago
Anthropic Hires Weapons Expert to Set Claude Safety Boundaries
Anthropic makes Claude, an artificial intelligence system.
The company is hiring a weapons expert to help set rules for what Claude can do.
The expert will decide which military-related requests are safe and which could help build weapons.
Some technologies, such as robots and navigation systems, can be used both for civilian purposes and weapons.
The expert will test how Claude might be misused.
They will also work with engineers to create protections that block harmful uses.
The job includes concerns about systems that can choose and attack targets without human approval.
Anthropic says it is doing this because its AI systems are becoming more capable.
Anthropic is hiring a Policy Design Manager specializing in conventional weapons.
The position will help distinguish legitimate engineering research from requests that could support weapons development.
The successful candidate will create threat models and evaluations for Claude’s potential misuse.
The role includes converting policy decisions into model guardrails, detection systems and enforcement tools.
Anthropic is offering an annual salary of $245,000 to $285,000 for the position.
- Who
- Anthropic is hiring a weapons expert for its Safeguards organisation.
- What
- The expert will help define and enforce rules for Claude’s assistance with conventional weapons-related work.
- Where
- When
- The hiring comes as Anthropic expands its AI safeguards; the article says the company published related research last week.
- Why
- Anthropic wants to reduce the risk that increasingly capable AI systems could be misused for weapons development or other high-risk activities.
Key facts
- Job title
- Policy Design Manager, Conventional Weapons
- Organisation
- Anthropic’s Safeguards organisation
- Annual salary
- $245,000 to $285,000
- Required expertise
- Practical knowledge of weapons systems, including experience at a defence research laboratory, government research agency or weapons company
- Preferred knowledge
- Machine learning, robotics, autonomy, guidance and navigation, sensors, aerospace or embedded software
- Main responsibilities
- Developing threat models and evaluations and helping implement guardrails, detection systems and enforcement tools
- Specific high-risk area
- Autonomous systems capable of selecting and engaging targets without human authorisation
Sources
Related news
Anthropic Reports Claude Used for Weapons and Influence Operations
Anthropic Blocks AI Misuse Linked to Potential Bioweapons Research
Anthropic Says Yemen Cell Used Claude for Missile Guidance
Appeals Court Upholds Pentagon’s Anthropic Supply Chain Risk Designation
Anthropic Regains Favor With Trump Administration, Commerce Chief Says
Anthropic CEO Urges Slower AI Development Amid Misuse Fears
Anthropic’s AI Consciousness Debate Reaches Religious Leaders and Critics
Anthropic Considers November IPO Amid Valuation and Loss Concerns
Anthropic Consults Religious Scholars About AI Morality and Consciousness
xAI Reportedly Considers Four Combined Grok and X Plans
OpenAI Alleges Moonshot AI Led Campaign to Copy Models
