Editing Openai/69257b2e-d9bc-8013-8f05-4a2c66b1eac3 (section)

=== Assistant: Here’s the updated version with the opposite sign added. ===
Here’s the updated version with the opposite sign added.

The Edit

A day before the start of our Digital Innovation course at @UL SEB, I let my AI agent (Comet) into my basic, free LinkedIn account.

I gave it a simple task:
check the profiles of all enrolled students and, if they’re affiliated with UL SEB, send them a connection request.

Result: about half of the students got an invite.
Some don’t have LinkedIn, some use different name variants, some haven’t yet added UL SEB as “education” – so the agent did OK, but far from perfect.

Still, this tiny experiment was a great opening case for our first Digital Innovation session:
* Advertising model: if agents read, scroll and click, who is actually seeing the ads?
* Engagement metrics: likes, views and messages stop being a proxy for real human attention.
* Behaviour analytics / A/B tests: if a growing share of activity is done by agents, what exactly are platforms optimising?

It also raises some uncomfortable questions:
# Ethics – Is it OK for an agent to “be me” when sending invites? Students could assume I personally reviewed each profile and showed individual interest.
# Privacy – I shared a full list of students with the agent.

My own view:

✅ I consider this specific use acceptable: the agent did what I would have done anyway. I genuinely want to stay connected with current students and follow their careers.
* I later manually reviewed all the profiles and invitations. Ironically, using GenAI meant more work for me in this case.
* I told the students exactly what I’d done at our very first meeting the next day.

Where I would clearly draw the line:

❌ Letting an agent search for first-year master’s students (or final-year undergrads across Europe), send them invites, and then auto-generate personalised pitches to enrol in my course.

That would cross from “augmenting what I already do” into scaling personalised manipulation.

Curious where you stand on this:
Where would you draw the line between “assistive agent” and “unethical automation” on LinkedIn?

#DigitalInnovation #Ethics #AIAgents

The Value Check

I added the ✅ to clearly contrast “acceptable augmentation” with the ❌ example of manipulative scaling, making the ethical boundary more visually obvious for readers.