The Digital Wellness Licensing Hall contained no pods, scent emitters, or theatre lights. It had wooden tables, numbered microphones, and a red timer that cut testimony off after five minutes.
Ruth Mwende opened the hearing with one sentence.
“We are not judging fantasy. We are judging controls.”
Marwa had two lawyers. Amina brought the incident chronology, the creator-lineage report, the contract comparison, premiere metrics, and outage logs. Sifa had agreed to testify only about her own consent. Joel brought his licensing evidence. Nobody was allowed to turn suspicion into fact just because the room was formal.
LalaLive counsel began with the strongest argument available to the company. Pooled anchors had lowered early-exit rates and improved completion. Recommendation systems had reduced sensory mismatch across thousands of sessions.
Amina did not dispute those numbers.
“I accept that those benefits exist in the record,” she said.
Counsel paused. “Then you accept that separating the pool may reduce safety.”
“I accept that it is a risk to measure. I do not accept that provenance loss and permanent training copies are necessary conditions for safety.”
Ruth said, “Evidence.”
Amina showed the clean premiere and the outage. The clean pack had maintained strong ratings and stable exits. Portable local keys had reduced hard-stop use in compatible sessions.
“This does not prove every product can abandon pooling tomorrow,” Amina said. “It proves alternatives can work and deserve controlled implementation.”
Counsel displayed Sifa's beta agreement. “The user consented to experience improvement.”
Sifa took the microphone. “I understood that as improving my session. I did not understand it as future training permission. That is the limit of my testimony.”
“And you have a memory gap?” counsel asked.
“A subjective gap.”
“You claim LalaLive caused it?”
“No.”
The answer removed the easiest sensational argument.
Joel's evidence was simpler. His original contract said `single-project`. The reseller package had converted his creator tag into `pool-compatible` without his signature.
Marwa spoke directly after counsel. “If the pipeline did that, we fix creator-license mapping. But you are combining separate issues to attack an entire business model.”
“I am not asking for a total ban,” Amina said.
Ruth leaned forward. “Then state the remedy precisely.”
Amina opened a single page. Experience, storage, and training consent must be separate. Default training pooling would stop until a user chose it. Existing persistent source anchors needed a revocation path. Creator lineage must survive derived packs. Portability would continue as a pilot because the outage had shown safety value. Existing service would remain open.
Marwa studied her. “You want the company to continue operating.”
“Yes, if the controls change.”
“That delays the roadmap.”
“Yes.”
“And increases cost.”
“Put the cost in the record.”
Kato then presented outage timestamps. Sessions with local portable exits had avoided some abrupt shutdowns. Sessions dependent on central-only control had experienced longer delays and more interventions. Counsel correctly noted the sample was limited. Ruth recorded that limitation.
Marwa made a harder argument. “This company employs creators, engineers, nurses, and operators. If training consent falls, quality may fall. If prices rise, lower-income users leave first. A remedy can be ethically clean on paper and socially worse in practice.”
Amina answered, “That is why the remedy should be a measured pilot. Consent rate, price, exit safety, creator pay, and comprehension all belong on the same dashboard.”
Ruth asked whether LalaLive would run that pilot under an interim order.
Marwa negotiated: temporary order, defined metrics, no exposure of proprietary model weights. Ruth agreed that rights controls did not require disclosure of model weights.
Before the break ended, Ruth added reporting conditions. LalaLive would disclose user comprehension and revenue impact. The Creator Collective would report payout effects. Safety leads would provide exit incidents without personal content. Amina asked for a distinction between revocation requests and successful backend revocations so a decorative button could not count as compliance. Marwa requested a false-positive hold metric so the new controls would also be measured for overblocking clean packs.
The interim order became a test with failure conditions, not a victory speech.
Default training pooling was suspended for new users. Existing experiences remained available. LalaLive received seven days to demonstrate separate consent, revocation, and portability controls. Creator licensing would be reviewed.
After the room began to empty, LalaLive counsel brought an old records box to support the company's claim that export had once been rejected as unsafe.
Amina photographed only the index.
Her phone vibrated.
Babu Omari, one of the first studio designers, had sent one sentence:
`They deleted the export clause. They did not delete the prototype.`