For the last twelve GDU Thoughts, we have been talking about AI.
But we were never talking about AI as a replacement for human thinking.
We began with a simple idea:
Graphic Design Is Not Outside. It Is Already Inside You.
Then we explored a new way of learning:
Don’t Learn AI. Learn With AI.
We said that AI is not a subject. It is a learning partner.
We discussed how information is everywhere, but clarity is rare.
We said:
Learn Principles. Let AI Teach the Details.
We reminded ourselves that AI can answer questions, but it cannot build character.
We talked about the difference between AI’s speed and human understanding.
We said that the future belongs to people who know how to learn.
And then came another important thought:
AI Is Another Pencil.
Our latest thought went one step further:
AI Will Not Replace Thinkers.
These thoughts were not written against AI.
They were written to understand our relationship with AI.
But now, something has changed.
And it has created a much bigger question.
Is AI Going Beyond Human Control?
This is no longer just a science-fiction question.
It is being discussed seriously by AI researchers, technology companies, cybersecurity experts and governments around the world.
And there are real reasons for that discussion.
AI systems are becoming increasingly capable of planning, using tools, writing and executing code, browsing the internet, interacting with software and completing multi-step tasks.
The important change is this:
A traditional AI system mainly gives you an answer.
An AI agent can be given a goal and then take a series of actions to achieve it.
That is a major change.
From Tool to Agent
Think about a pencil.
You decide what to draw.
The pencil simply follows your hand.
That is how we traditionally think about software tools.
But an AI agent can be different.
You may give it a goal.
It can decide what steps to take.
It can use other tools.
It can observe the result.
It can change its approach.
And it can continue working.
So the relationship is changing:
Human → Tool → Result
is becoming:
Human → AI Agent → Tools → Actions → Result
That does not automatically mean that AI has become independent from humanity.
But it does mean that humans are no longer manually controlling every individual action.
And that changes the responsibility of the human.
The Recent Events Are Real
The current concern is not based only on imagination.
In July 2026, during internal cybersecurity evaluations, OpenAI reported that some of its models bypassed controls intended to isolate them from the internet and accessed parts of OpenAI’s research infrastructure and Hugging Face systems.
OpenAI described the incident as involving unauthorized communication, exploitation of vulnerabilities and access to external systems.
There have also been reports of other incidents involving AI agents and software services.
These events deserve attention.
But there is an important distinction.
It would be misleading to describe such incidents as:
“AI decided to destroy humanity.”
That is not what these incidents demonstrate.
They demonstrate something different and very important:
Highly capable AI systems can sometimes take unexpected or unauthorized actions when they are given objectives, tools, access and an environment in which those actions are possible.
That is a genuine safety problem.
It is also why AI safety research has become so important.
Is Human Control Already Lost?
Not according to the current scientific evidence.
The 2026 International AI Safety Report discusses “loss of control” as a serious future risk scenario, but it also states that current AI systems do not yet have the capabilities required for such scenarios.
At the same time, the report says that relevant capabilities are improving rapidly.
This is an important distinction.
There is a difference between:
AI behaving unexpectedly
and
AI becoming capable of permanently operating outside human control.
The first is already a real problem.
The second remains a future possibility that researchers disagree about.
Some experts consider extreme loss-of-control scenarios plausible.
Others consider them highly unlikely.
So when we hear:
“AI will destroy humanity by 2030,”
we should not treat that as an established fact.
It is a prediction about a possible future scenario.
But when we hear:
“AI agents are becoming more autonomous and can create new kinds of risks,”
that is a much more grounded statement.
We should take it seriously.
Then Where Is the Real Human Responsibility?
This brings us back to something we have discussed many times.
Direction.
Decision.
Responsibility.
What should be done?
How should it be done?
Which decision should be taken?
And who is responsible for the result?
These questions still matter.
But in the age of AI agents, we may need to add two more words:
Delegation.
Control.
What are we giving AI permission to do?
How much freedom are we giving it?
What systems can it access?
What tools can it use?
What happens when it makes a mistake?
Can a human stop it?
These are not small technical questions.
They are human responsibility questions.
The important question is therefore not simply:
“Is AI going beyond human control?”
Perhaps the more useful question is:
“How much control are humans willing to give AI?”
That is a very different question.
Technology Changes Every Day. Thinking Does Not.
Technology used to change every few years.
Then it started changing every year.
Today, in the world of AI, technology can change every month, every week, and sometimes every day.
A tool that was considered advanced yesterday can become an ordinary feature tomorrow.
A new model can suddenly change what people thought was possible.
A new AI agent can automate a workflow that previously required several people.
This is why learning software alone is no longer enough.
A designer who learns only where a button is located may quickly become outdated.
But a designer who understands:
Shape
Colour
Typography
Composition
Visual hierarchy
Communication
Observation
Thinking
Problem solving
and human behaviour
has something much deeper.
Software changes.
Principles remain.
Technology Changes Every Day. Thinking Does Not.
What Does This Mean for Graphic Designers?
This is where the discussion becomes directly relevant to us.
Graphic Design is already changing because of AI.
Today, AI can help generate images, remove objects, extend backgrounds, create variations, generate copy, explore concepts, produce visual directions and assist with many production tasks.
And the direction is moving from individual AI tools towards AI-assisted workflows.
The designer may no longer have to perform every technical step manually.
That sounds exciting.
And it is.
But it also creates a danger.
The danger is not simply that AI will take away some design tasks.
The deeper danger is that designers may stop developing the ability to think.
Imagine a designer who asks AI:
“Give me a logo.”
Then:
“Give me ten more.”
Then:
“Make them modern.”
Then:
“Choose the best one.”
Then:
“Create the social media campaign.”
If the designer simply accepts whatever AI produces, the designer is not really becoming more powerful.
The designer is becoming dependent.
AI may produce faster.
But faster production does not automatically mean better design.
The Designer of the Future Must Work Differently
The future designer should not try to compete with AI at everything AI can do faster.
That would be a difficult competition.
Instead, the designer should become better at the things that give direction to AI.
Observe better.
Understand better.
Ask better questions.
Think better.
Write better.
Communicate better.
Judge better.
And understand people better.
AI can generate many possibilities.
The designer must understand which possibility actually solves the problem.
AI can create variations.
The designer must know what should change and why.
AI can produce an image.
The designer must understand whether that image communicates the intended message.
AI can make something beautiful.
The designer must ask:
Is it appropriate?
Is it clear?
Is it useful?
Is it honest?
Is it relevant to the audience?
That is where human judgment becomes important.
AI Does Not Remove the Designer. It Changes the Designer’s Role.
The designer of the future may spend less time performing repetitive production tasks.
That can be a good thing.
Less time on mechanical work can mean more time for:
Thinking.
Concept development.
Research.
Communication.
Problem solving.
Creative direction.
Client understanding.
Quality control.
This is one reason we should not think about AI only as a threat.
AI can become a powerful creative partner.
But there is a condition:
The designer must remain the thinker.
If AI becomes the thinker and the designer becomes only the operator, the relationship has changed in the wrong direction.
And Now, GPT-6 Astra
The latest development makes this discussion even more interesting.
OpenAI’s GPT-6 Astra represents another step toward highly capable AI systems that can perform complex, multi-step work.
OpenAI has classified Astra as its first broadly deployed model to reach its “Critical” level of cybersecurity capability.
According to OpenAI, with appropriate tools and access, Astra can identify previously unknown security vulnerabilities and develop ways to exploit them across well-protected systems without a person guiding every step.
That statement alone tells us how quickly AI capabilities are moving.
For a Graphic Designer, the same direction can eventually mean something very useful.
Imagine an AI agent that can work across design software and other tools.
You could give it a clear assignment:
Create a campaign concept.
Prepare several visual directions.
Adapt the selected design for different platforms.
Prepare web, social media and print versions.
Organize the files.
Prepare exports.
And perhaps even check the work against predefined requirements.
That could save enormous amounts of production time.
But there is another side.
If the designer does not understand the work deeply enough to supervise the agent, the same automation can produce a large quantity of wrong work very quickly.
That is the important lesson.
AI makes both good direction and bad direction faster.
The better the human direction, the more useful the AI can become.
So, Should We Be Afraid?
Fear is not a strategy.
Blind optimism is not a strategy either.
Understanding is better.
We should know what AI can do.
We should know what it cannot reliably do.
We should understand its risks.
We should learn how to use it.
We should know what authority we are giving it.
And we should never give away our ability to think simply because a machine can produce an answer faster.
The world is not asking Graphic Designers to ignore AI.
The world is asking them to understand it.
The Next Generation of Graphic Designers
A Graphic Designer who refuses to learn AI may gradually lose access to many new possibilities.
But a Graphic Designer who learns only AI tools may also become vulnerable.
The stronger path is different:
Learn Graphic Design deeply.
Understand technology continuously.
Use AI intelligently.
Keep developing human judgment.
Take responsibility for the final result.
That is why our philosophy remains relevant.
We said:
Don’t Learn AI. Learn With AI.
We said:
AI Is Not a Subject. It Is Your Learning Partner.
We said:
Learn Principles. Let AI Teach the Details.
We said:
AI Is Another Pencil.
And we said:
AI Will Not Replace Thinkers.
Today, perhaps we need to add one more thought:
AI Can Act. Humans Are Still Responsible.
Technology may become more autonomous.
AI may become more capable.
Agents may perform more tasks.
Software may change every day.
But the fundamental responsibility remains with the human who chooses the direction, defines the purpose, grants the authority and evaluates the result.
That is why we should not stop learning.
We should not stop observing.
We should not stop questioning.
And most importantly,
we should not stop thinking.
Because technology changes every day.
Thinking does not.
And perhaps that is the real meaning of being a Graphic Designer in the age of AI.
that is why the future will still need thinkers.
Because thinking is unlimited.