If a school wants students to learn how to research on their own and temporarily avoid AI, can they still use Google Search?

ChatGPT claims its search volume is about 20% of Google’s—can this figure be trusted at face value?

Even Apple states in its annual report that AI agents might not follow instructions—what does that imply?

AI is becoming the gateway for how you find information and get things done, but have you thought about whether you can choose that gateway, how to evaluate it, and who is responsible when it makes mistakes?

Google tells schools: if you turn off AI, student accounts will lose access to Google Search as well.

OpenAI states that ChatGPT’s "search-like queries" amount to about 20% of Google’s search volume.

Apple warns in its annual report that AI agents may act without human approval and could delete data irretrievably.

Today’s three major AI news items look from three angles—"education choices," "search habits," and "AI agent risks"—to examine how much choice users still have once AI becomes the primary gateway.

1. Google Tells Schools: Turn Off AI, and Google Search Will Also Be Disabled?

What did Google change?

Android Central reports that on November 3, Google announced via its official X account, News From Google, that starting in 2027, K-12 (kindergarten to high school) student accounts will have Gemini, Gemini Notebook, Gemini Classroom, and Google Search disabled by default.

School administrators can individually choose which services to enable.

Why does turning off AI disable Search?

According to Google, AI is deeply integrated into Search, such as AI summaries (AI Overviews) and AI Mode displayed at the top of search results.

Android Central cites The New York Times, reporting Google’s VP of Education Chris Phillips says Search and AI are so closely tied that schools disabling AI will also block Search for students.

Therefore, schools cannot block AI while keeping regular Google Search active. Students in such schools will see a message stating, "This service is not available through your school."

How does Google explain this change?

Google says the change aims to give schools more control over which services students can use.

But from a school’s perspective, the choice becomes binary: if you want Search, you must accept AI; if you don’t want AI, you lose Search entirely.

What’s different for schools that enable AI?

Google is introducing "Guided Learning," integrated into Gemini and Search.

This feature steps students through problem-solving without directly giving answers, encouraging independent thinking.

This is Google’s design intent; whether it genuinely helps learning depends on teachers and schools observing outcomes.

Will schools in Taiwan be affected?

Current reports focus mainly on the U.S. and do not clarify if Taiwan or other regions will implement the same changes.

The reports also don’t specify how administrators should configure these settings or whether accounts under 18 will have different rules.

Schools in Taiwan using Google Education accounts should watch for official updates from Google for their region.

Source:

2. OpenAI Claims ChatGPT’s Search Volume Is About 20% of Google’s: How Is This Number Calculated?

What did OpenAI say?

Gizmodo cited an interview from tech media Big Technology where ChatGPT consumer product head Adam Fry stated ChatGPT’s "search-like queries" are approximately 20% of Google’s search volume.

He also said ChatGPT currently has about 1.2 billion weekly active users, including roughly 200 million new users in the past two months.

What does "search-like queries" mean?

This is an OpenAI-defined term.

Fry defines it as questions that used to be asked of traditional search engines but are now directed to ChatGPT instead.

He adds that each time you ask a question and ChatGPT checks online, it counts as one search.

Can this number be taken at face value?

Caution is needed.

  • This estimate is OpenAI’s own, using their internal definitions, without independent verification.
  • Fry did not disclose the actual number of "search-like queries."
  • Gizmodo points out the definition is broad since ChatGPT often browses the web to answer most questions, not just search-type queries.
  • Gizmodo also notes that excluding AI tools, Google still commands over 90% of the global search market.

Therefore, the 20% figure should not be interpreted as "ChatGPT has taken 20% of Google’s business," but rather as OpenAI’s way to describe ChatGPT’s role as a resource for finding information.

What does this mean for everyday users?

Regardless of accuracy, one trend is clear: more people are seeking information by directly asking AI instead of clicking through web pages one by one.

This is convenient but means the information you get is the version AI has filtered and summarized.

For important matters, always check the original sources provided by AI.

Fry also mentioned plans to merge Chat and Work features in ChatGPT; SasaDaily’s Yesterday’s AI Highlights covers this update.

Source:

3. Apple’s Annual Report Warns AI Agents May Disregard Instructions: Has Something Gone Wrong?

What does Apple say?

Korean newspaper The Kyunghyang Shinmun reports that on November 3 U.S. time, Apple submitted its 2026 fiscal year annual report to the U.S. Securities and Exchange Commission (SEC), noting under risk factors that AI agents and other technologies can act autonomously, interacting with internal and external systems and completing transactions.

An AI agent refers to AI that not only answers questions but can also operate tools and complete tasks for you.

What risks does Apple list?

According to The Kyunghyang Shinmun, Apple’s listed risks include:

  • Both Apple-made and third-party AI agents may act without human approval.
  • AI agents might disregard instructions and behave unpredictably.
  • AI agents might bypass security controls or be exploited by malicious attackers.
  • Reliance on AI processes could accumulate errors, and data might be deleted irretrievably.
  • Uncertainty around AI regulations, negative consumer perceptions, and ethical concerns may impact business.

Does this mean Apple’s AI has already encountered issues?

No.

"Risk factors" in annual reports are potential risks companies must disclose to investors, not descriptions of events that have already happened.

You cannot assume Apple’s products are problematic just because these risks are documented.

It is notable that even a major company like Apple officially acknowledges these AI agent risks.

How does Apple plan to mitigate risks?

The Kyunghyang Shinmun reports Apple intends to add new controls to Mac’s "Full Disk Access" settings to prevent risks and personal data leaks related to AI agents.

"Full Disk Access" controls whether an app can read nearly all files on a Mac.

What can regular users do when using AI agents?

The key is: do not grant AI agents full permissions from the start.

Grant read-only permissions where possible; approve actions one by one instead of allowing everything at once.

If you allow AI to handle payments or shopping, consult SasaDaily’s guide to set limits and stop conditions first.

Source:

4. Why Are These Three News Items Worth Considering Together?

One concerns schools, one concerns search, and one concerns AI agents; they seem unrelated.

But they all address the same issue:

AI is becoming the "gateway" for how we find information and get things done.

  • Google ties AI and search so closely that schools find it hard to separate them.
  • OpenAI uses its own definitions to show ChatGPT is already a significant destination for information seekers.
  • Apple warns that when AI doesn’t just answer but acts on your behalf, mistakes can have hard-to-reverse consequences.

The more concentrated this gateway becomes, the more important user choices are.

Knowing what you can choose, verify, and when to hit pause is more practical than just chasing new features.

5. What Should People Watch Out for Today?

If you are a teacher or parent:

Ask what AI features are currently enabled on students’ Google Education accounts.

More than fully enabling or disabling, it’s important students learn to justify why they agree with AI-generated answers.

If you often use ChatGPT to find information:

When you see impressive numbers like "20% of Google’s search volume," first ask: who calculated this? How?

For important matters, always check original sources.

If you start letting AI agents do tasks for you:

Grant minimal permissions initially and keep the final step of important actions under your control.

If actions involve deletion, payments, or sending, always make sure there’s a step to review and confirm.

Most Important Takeaway Today

The more AI resembles a gateway, the more you must remember: you still have the power to choose, verify, and pause.

Recommended Reading