[My previous response to similar concerns]({{<relref "/notes/on-enforcing-https.md">}}) is relevant. To elaborate:
If nothing prevents bad behavior from an ISP, and it has happened before, then you should assume it's happening. This extends to injecting JavaScript apps into insecure connections.
- [Infrastructure likely belonging to the Great Firewall of China tampers with Baidu analytics to DDoS GitHub (2015)](https://www.eff.org/deeplinks/2015/04/china-uses-unencrypted-websites-to-hijack-browsers-in-github-attack)
- [Comcast continues to inject its own code into websites you visit (2017)](https://thenextweb.com/news/comcast-continues-to-inject-its-own-code-into-websites-you-visit)
- [How is my ISP able to inject into this webpage? (2019)](https://security.stackexchange.com/questions/221260/how-is-my-isp-able-to-inject-into-this-webpage)
- [Optimum ISP is MITMing its customers (2023)](https://lukerodgers.ca/2023/12/09/optimum-isp-is-mitming-its-customers/)
Unless you trust every hop from your browser to the destination server (and back), assume anything unencrypted can and will be inspected (and potentially tampered with). Encrypt everything you can.