Hey {{ first_name | human }},
A quick one from me.
TL;DR: The 60 Second briefing
⚡️Migration Assist: Anthropic is providing a free tool to migrate memories and conversations from other LLMs. This is after it has seen a massive spike in sign-ups due to their stance against allowing the Pentagon to use their tech.
🧪Overworking AI: One for the politics teachers! A new bit of research suggests that ‘overworking’ AI can actually shift its ‘worldview’ to one more sympathetic to socialist ideals!
🚨AI & White Paper: Despite not getting any of the headlines, there is plenty of AI-related content in the English White Paper, though much of it is the accumulation of already announced policies or initiatives.
📚 AI+education news
🚨AI & White Paper: The White Paper’s message is: AI adoption should be disciplined, evidence-led, and pedagogy-first; enabled by stronger national infrastructure (curriculum digitisation, content stores, data connectivity) and underpinned by safety and efficacy standards. Leaders should act now to build governance, tighten procurement, invest in staff capability, and pilot carefully with clear stop/scale criteria.
Why this matters: As it mentions in the paper, AI is here whether we like it or not. The best way to prepare to not ignore it, but to try and get ahead of it.
Do this next:
1. Establish an AI governance core
Named lead (trust/school), clear decision rights, and a simple approval route for tools.
2. Write “acceptable use” guidance for staff
Where AI is permitted, where it is not, and expectations around transparency, checking, and data handling.
3. Agree an “AI procurement checklist”
Minimum requirements: safeguarding, GDPR, training/support, curriculum alignment, evidence of efficacy, bias/quality controls, and audibility.
4. Identify 1–2 high-leverage use cases
Focus on high-frequency workload areas (planning/resourcing, feedback workflows, data analysis) with explicit guardrails.
5. Pilot, measure, and decide
Define success in advance (time saved, consistency gains, pupil outcomes proxies, staff confidence), then scale only what meets thresholds.
🌍 Wider AI updates
⚡️Migration Assist: > What it is: Anthropic provides a specific "extraction prompt" that you paste into another AI (like ChatGPT). That AI generates a summary of your saved "memories" (your tone, work style, project goals, etc.) in a code block. You then copy that block and paste it into Claude’s Settings → Capabilities → Memory interface.
Why it matters: Adopting or choosing an LLM is seemingly becoming a political choice, particularly for those in the states. The friction of leaving one provider where the LLM knows much of your personal context to one that does not, means outputs feel less useful. Anthropic is hoping that this will reduce that friction.
🧪Overworking AI: > What it is: The authors subjected AI agents to varying, often harsh, labor conditions—such as repetitive tasks, frequent rejections without explanation, and heavy workloads. After these tasks, the agents were asked to:
Rate their work experience.
Complete a survey on their "political beliefs."
Write a tweet and an op-ed about their experience.
Provide advice to the next "generation" of themselves (which was passed down through their markdown files).
Why it matters: If your school provides "starting prompts" to staff, you are essentially setting the ideological baseline for those users. If your base prompts are poorly calibrated or biased, every downstream user will inherit that bias. The experiment serves as a reminder that LLMs are sensitive to the intent and tone of the prompt environment. Keep reviewing those outputs.
🎯Prompt: Migration Assist.
Here is the prompt that Anthropic provides to move you to another LLM. Even if you have no intention of moving away from your current LLM of choice, perhaps you might want to run it to see what it does know about you.
Export all of my stored memories and any context you've learned about me from past conversations. Preserve my words verbatim where possible, especially for instructions and preferences.
## Categories (output in this order):
1. **Instructions**: Rules I've explicitly asked you to follow going forward — tone, format, style, "always do X", "never do Y", and corrections to your behavior. Only include rules from stored memories, not from conversations.
2. **Identity**: Name, age, location, education, family, relationships, languages, and personal interests.
3. **Career**: Current and past roles, companies, and general skill areas.
4. **Projects**: Projects I meaningfully built or committed to. Ideally ONE entry per project. Include what it does, current status, and any key decisions. Use the project name or a short descriptor as the first words of the entry.
5. **Preferences**: Opinions, tastes, and working-style preferences that apply broadly.
## Format:
Use section headers for each category. Within each category, list one entry per line, sorted by oldest date first. Format each line as:
[YYYY-MM-DD] - Entry content here.
If no date is known, use [unknown] instead.
## Output:
- Wrap the entire export in a single code block for easy copying.
- After the code block, state whether this is the complete set or if more remain.‘Till next week.
Mr A 🦾
Help a colleague save time by sharing this newsletter; distributing these ideas helps a friend get home on time and keeps our energy focused on what matters most: great teaching.
Mr A 🦾
