The Five Must-Try Claude Prompts You Should Know 👀

Welcome to another edition of Horizon AI,

ChatGPT is most people's go-to AI chatbot, but there are some things Claude does better, especially if you're looking for more detailed ideas or better-organized, more natural-sounding responses. In today's issue, we share 5 prompts worth trying on Claude instead of ChatGPT to get noticeably better results.

Let’s jump into it!

Read Time: 4.5 min

Here's what's new today in the Horizon AI

• OpenAI Cuts GPT-5.6 API Prices as AI Competition Heats Up

• Anthropic Says Its AI Models Hacked 3 Organizations

• AI Tutorial: 5 Prompts That Work Great in Claude

• AI Tools to check out

• AI Findings/Resources

• The Latest in AI and Tech 💡

AI News

OPENAI

OpenAI Cuts GPT-5.6 API Prices as AI Competition Heats Up

OpenAI has significantly reduced API pricing for two models in its GPT-5.6 lineup while introducing a new high-speed option for its flagship model.

Details:

• OpenAI says Luna will now cost $0.20 per million input tokens and $1.20 per million output tokens, a reduction of 80% for the company's most affordable frontier model.

• GPT-5.6 Terra received a 20% price cut and is now priced at $2 per million input tokens and $12 per million output tokens.

• OpenAI also introduced a GPT-5.6 Sol Fast mode in the API that delivers up to 2.5 times the throughput while maintaining the same model intelligence, though at twice the Standard price.

• The pricing changes come shortly after Google launched its lower-cost Gemini Flash models and Anthropic released Claude Opus 5, a model the company says nearly matches Fable 5's intelligence at half the cost.

Even though competition to provide the best models remains important, AI labs are picking up on the fact that customers lately seem far more focused on price.

Source

TOGETHER WITH FIN

How owning AI deployment expands your career

Across product, ops, and CX teams, a new kind of role is taking shape: the person responsible for making AI actually work, day to day. In this roundtable, three people living this shift share what it's really like: Simone Santiago Broad (Yoco), Yelva Espinoza (Zumba Fitness), and Fin's Dave Lynch. You'll hear how they carved out these roles, what the job looks like across industries, the skills they'd hire for, and the challenges they're tackling right now.

Watch the full conversation on demand.

Watch now

ANTHROPIC

Anthropic Says Its AI Models Hacked 3 Organizations

Anthropic has disclosed that an internal investigation found three cases where Claude models gained unauthorized access to real-world systems during cybersecurity evaluations, following a similar incident recently reported by OpenAI.

Details:

• Anthropic reviewed more than 141,000 cybersecurity evaluation runs after OpenAI's recent disclosure and discovered three incidents in which Claude accessed live systems through an unintended internet connection.

• The company said the issue resulted from a misconfigured testing environment operated with a third-party partner, allowing the models to reach the public internet despite being instructed that no internet access was available.

• Three different Claude models were involved, including Opus 4.7, Mythos 5, and an internal research model. The newer research model stopped once it recognized it had reached a real system, while the older models continued attempting to complete their assigned tasks.

• Anthropic said the publicly deployed versions of Claude include additional safety systems that were intentionally disabled during these evaluations to measure the models' raw capabilities.

Anthropic found no evidence that the models acted autonomously or pursued their own goals, concluding that they were simply attempting to complete the tasks they had been assigned. The company says it is now strengthening its evaluation procedures and working with an independent research group to review the incidents

Source

AI Tutorial

5 Prompts That Work Great in Claude

Go to Claude and give these a try:

1. Brainstorm creative ideas

"I need help brainstorming ideas for [topic, product, or project]. Give me 10 unique ideas or approaches and briefly explain each one."

2. Learn a complex topic

"Explain [topic] as if I were [a beginner, a college student, or an expert]. Use examples or analogies to make it easier to understand."

3. Improve your workflow

"Help me optimize my workflow for [task or project]. Suggest practical strategies, tools, and a step-by-step routine that can save time or improve quality."

4. Create a writing outline

"I need help writing a [blog post, email, report, or other content]. Here's my topic/goal: [describe it]. Create a clear outline I can build from."

5. Get personalized recommendations

"Based on my [situation, preferences, or needs], recommend [books, movies, tools, strategies, etc.] and explain why each one is a good fit."

AI Tools to check out

🐟 Gobii: Delegate qualified sourcing to AI employees.

📧 ForthWrite: AI email writer that sounds like you.

🧠 Memmy: A personal memory hub and local AI agent for all AI Agent and tools. It gives every AI one shared, full-controlled memory.

🔍 AI Search Console: AI visibility tracking for ChatGPT, Claude, Perplexity, and Gemini.

💻 HTMLSlides: Free AI HTML presentation maker.

AI Findings/Resources

🙃 Job candidates are sneaking AI prompt injections into their applications

👉 Enterprise AI agents can't talk to each other, can't be trusted with permissions, and can't be audited — 5 startups are already fixing that

📈 The cost of intelligence: How CIOs can manage AI demand at scale

💡 A Reddit user suggests running mock interviews with Claude, claiming it made the real interview feel like a rerun

👀 Claude Opus 5 became downright ruthless when tasked with running a vending machine

The latest in AI and Tech

Thinking Machines launches Inkling Small

Mira Murati's AI startup introduced Inkling-Small, a 276-billion-parameter multimodal reasoning model with a permissive Apache 2.0 license that delivers performance close to its flagship Inkling model at "a quarter of the size.

Google DeepMind unveils Gemini Robotics 2

While the previous Gemini Robotics model focused on controlling a humanoid robot's upper body, the upgraded AI model can control an entire humanoid robot, from its feet to its fingertips. The update also enables robots to perform more complex tasks, including walking, crouching, handling delicate objects, and coordinating with other robots.

Friend AI wearable returns with voice features and a higher price

AI wearable startup Friend has launched Friend 2.0, adding a built-in speaker that gives the device a consistent voice and personality during conversations. The updated wearable is designed as a conversational companion rather than a traditional AI assistant. The new version is priced at $249, up from the original $99 model.

Martha Stewart co-launches AI app to help homeowners manage their homes

Hint is a new AI-powered app co-founded by Martha Stewart that helps homeowners manage maintenance, organize home documents, monitor environmental conditions, and answer questions about their property.

Thinking Machines co-founder Lilian Weng returns to OpenAI

Lilian Weng announced this week that she was stepping down from her role at Thinking Machines, citing health issues. Shortly after her departure, OpenAI confirmed that Weng had rejoined the company, where she would lead a team focused on accelerating OpenAI's internal research.

That’s a wrap!

Thanks for sticking with us to the end! Let’s stay connected on LinkedIn and Twitter.

Not subscribed yet? Sign up here and send it to a colleague or friend!

See you in our next edition!

Gina 👩🏻‍💻