GPT-5 Chat Draws Fire | Users Claim Follow-Up Issues Persist

A rising wave of complaints from users highlights ongoing problems with GPT-5 chat's instruction adherence. Many assert that despite explicit commands to avoid follow-ups, the AI continues to ask clarifying questions, frustrating users in a fast-paced digital landscape.

The Context Behind the Controversy

In recent weeks, several individuals have reported that their custom instructions—their attempts to enhance interactions with GPT-5 chat—are being ignored. Users have noted that even when they hard-code requests for no follow-ups, the model persists in injecting phrases like “Want me to?” and “Can you clarify?” This phenomenon was not as prevalent in previous versions, sparking concerns about instruction adherence.

Reproducible Issues

Users have tried various methods to address the problem with little success. One stated, "I’ve hard-coded a 'no questions, no follow-ups' line at the top of my instructions the toggle does nothing." Another reiterated frustrations, saying, "GPT-5 chat still tacks on 'Want me to' every few replies." Given the feedback, many are seeking alternatives, reverting to the more compliant GPT-5 Thinking and GPT-4.1 models.

Patterns of User Sentiment

The collective feedback reveals a few key themes:

• Frustration with Activation Issues. Many users find the follow-up settings ineffective, noting that even double-layered instructions do not yield desired results.

• Better Performance in Previous Models. Users frequently compare the performance of GPT-5 chat unfavorably with GPT-5 Thinking and GPT-4.1, which are perceived to respect user constraints better.

• A Plea for Fixes. Comments reflect a desperate call for improvement: "This looks like a regression in instruction adherence on GPT-5 chat, not user error."

Key Points

• 🌟 Many users report that custom instructions to block follow-ups are ignored.

• 🔄 Complaints indicate a significant negative shift in performance compared to GPT-4.1 and GPT-5 Thinking.

• 📉 Frustrated users are switching back to older models in search of compliance.

The current situation raises a crucial question: How can developers ensure that AI models respect user preferences?

Epilogue

With ongoing feedback pushing for fixes, user frustration continues to build around GPT-5 chat’s capability to follow instructions. As conflicts like this unfold, the need for transparency and effective solution adaptation in AI technology becomes urgent. Users demand models that not only respond to inputs but also respect the boundaries they set.

Forecasting User Adaptation

There’s a strong chance that as frustration mounts, many people will shift away from GPT-5 chat entirely, gravitating back to previous models like GPT-4.1 and GPT-5 Thinking. Experts estimate around 60% of users could choose to abandon the AI for more reliable options if the follow-up issues persist. Developers may face mounting pressure to address these concerns, potentially leading to a rapid re-evaluation of their instruction adherence capabilities. If these fixes come through, we could see a renewed trust in the latest models; however, if users continue to feel ignored, the defection rate could increase even further, driving demand for more responsive AI solutions.

A Historical Echo in Communication Technology

In the late 19th century, the introduction of the telegraph dramatically changed how messages were sent, much like current advancements in AI communication. Early adopters struggled with unclear signals and frustrating delays, leading to a public outcry for improvement. Just as telegraph companies adapted their machinery and protocols in response to user complaints, it’s crucial for AI developers today to pay heed to the voices that feel overlooked. This disconnect between innovation and practical instruction could reshape how technology evolves, echoing history's insistence that respect for communication boundaries is paramount.